Features

Everything Enaibleu Sovereignty does today, grouped by where you meet it. Every entry here has shipped.

343 features across 15 modules

Last shipped 15 Nov 2026. What is next is on the roadmap; every release is in the changelog.

Enaibleu Sovereignty is a governed AI platform that runs on hardware you own: a framework of laws and roles, an autonomous workforce, persistent memory, local models, a crawler, a mailbox, an integration hub and gateway policies, with an audit trail under all of it. Items marked sovereign only work because the models, the data and the audit trail live on your own machine.

Agentic accountability

The runtime enforcement of the Sovereignty framework's Agentic Accountability law, so every AI action is attributable, permitted and reproducible.

  • Signed model registryPlatformsovereign

    Every model in the fleet carries a manifest with its weights digest, source, licence and promotion date. The router refuses a model whose digest drifted and health shows it.

    Shipped 27 Sept 2026
  • AI call lineage ledgerPlatformsovereign

    Each inference writes a hash-chained row with the agent, manifest, module, lane, resolved model, prompt and output hashes, tokens and latency, spooled locally when the store is away and sealed with a signed daily root that anyone can verify.

    Shipped 27 Sept 2026
  • Agent permission manifestsPlatformsovereign

    Each agent declares the tools, collections and hosts it may touch. The tool dispatcher checks the manifest on every cascade step and agent loop; an unknown agent is denied, a known agent without a manifest gets read-only tools, and every denial is logged.

    Shipped 27 Sept 2026
  • AI Act dossier generatorPlatformsovereign

    Produces the Annex IV technical documentation and a risk classification per AI use from the lineage ledger, the model registry and the routing configuration, as a versioned document per module.

    Shipped 27 Sept 2026
  • Immutable audit exportPlatform

    Hash-chains the audit history across the platform and the Robeeflow hub and exports it to write-once storage on a schedule, with a verifier anyone can run.

    Shipped 27 Sept 2026
  • Backup rehearsalOperations

    Scheduled PocketBase snapshots are restored into a scratch instance and a report says what came back, so a backup is proven rather than hoped for.

    Shipped 27 Sept 2026
  • Skills export as Agent Skills packagesPlatform

    Every framework skill can be exported as a standard Agent Skills package that any agent runtime can load. Outside skills come in only through an intake flow that records their origin.

    Shipped 27 Sept 2026
  • Sovereignty framework v11.1.0Platform

    The platform is governed by a versioned framework of laws, roles, skills, flows and templates with a proposal lifecycle and a linter. Version 11.1.0 adds the Agentic Accountability law, the Correspondent role, crawl and correspondence conduct and six runtime error codes.

    Shipped 27 Sept 2026Open
  • Work LogPlatform

    The platform records what people and AI agents did, through one function behind the API, the admin page, MCP tools, chat tools and cascade steps. Commits, changelog entries and each agent's daily activity are ingested automatically, and a period renders as a report with highlights, a timeline and decisions in Markdown, HTML and JSON.

    Shipped 27 Sept 2026Open
  • Sigil library and commandPlatform

    Parses, canonicalises, digests, seals, verifies and redacts Sigil documents, the open format in which the platform's records are written, so every record is checked by one implementation in the same way.

    Shipped 12 Nov 2026
  • Sealed framework documentsPlatform

    Every governed framework document has a sealed Sigil form beside the Markdown it keeps. People keep editing the Markdown; the sealed form is regenerated with it, so an outsider can verify the laws and decisions the agents work under.

    Shipped 12 Nov 2026
  • Signed agent requests and Sigil headersPlatformsovereign

    The platform's agents sign every outbound request with HTTP Message Signatures against a published key directory, and every API answer names its trace and the sealed record behind it. Each day of records is sealed with a signed, chained root and a timestamp, without a blockchain.

    Shipped 12 Nov 2026
  • Sigil verification on four channelsPlatform

    Anyone holding a record digest checks it on the dashboard and its API, in the AI chat, over MCP or between agents, and gets the same answer on each: the provenance, its place in the sealed day, the day's signature and chain, and the timestamp.

    Shipped 12 Nov 2026
  • Sigil open specificationPlatform

    The Sigil format is published openly with an implementer's guide, examples and test vectors, the text under CC BY 4.0 and the code under Apache 2.0, so others can implement and check it without us.

    Shipped 12 Nov 2026
  • Runtime reads the framework as SigilPlatform

    The runtime reads laws, manifests and the agents' context from the sealed form beside each document when it is in step and its seal verifies, and from the Markdown otherwise, saying why.

    Shipped 12 Nov 2026
  • External Sigil conformance suitePlatform

    230 cases across the four conformance levels let anyone check their own Sigil implementation against ours.

    Shipped 12 Nov 2026
  • Sigil export to OpenTelemetryPlatform

    Sealed records and days export as OpenTelemetry GenAI spans to existing observability tools, without the hashes of requests and answers.

    Shipped 12 Nov 2026
  • Public Sigil registryPlatform

    Our keys, the sealed days and the framework root are public at /.well-known/sigil, without a login; the records stay behind it.

    Shipped 12 Nov 2026
  • Signed API and chat answersPlatformsovereign

    Every API answer is signed with our published key over its status, body, trace, record and question; a chat answer carries the same signature in its footer and is checked in the browser.

    Shipped 12 Nov 2026
  • Checked facts in answersPlatform

    A value a tool returned, such as the time in a city or the temperature, is shown in the answer in the brand colour and linked to its source; a number the model wrote on its own stays plain.

    Shipped 14 Nov 2026
  • Decisions signed with a personal keyPlatformsovereign

    A decision on a proposal counts only when a person signs it with their own passkey or hardware key; the decision is sealed beside the proposal.

    Shipped 12 Nov 2026
  • Sign-in block listPlatform

    An administrator blocks an e-mail address from every way in, including Microsoft sign-in and API keys; open sessions end at once.

    Shipped 12 Nov 2026

Nexus workforce and cascades

Autonomous agents that run scheduled and event-driven work, file a report on every run and earn their standing in the open.

  • Nexus cascadesNexus

    Multi-step agent pipelines defined in YAML with solo, chain, event and loop types and http, agent, job, wait, geo lookup, database and OTEL logging steps. Event cascades watch any collection and fire when a record matches.

    Shipped 31 Mar 2026Open
  • Named agent personasNexus

    Police, Documenter, Janitor, Linter, Medic, Librarian, Enricher, Cipher, Hunter, Scout, Curator and Assessor each own their cascades under a distinct identity, so a report always says which agent did the work.

    Shipped 11 Jun 2026
  • Deal intelligenceNexus

    MEDDPICC scoring of a proposal or notes, then a second pass on the reasoning lane for the stakeholder map with stance and influence, risks with mitigations, a prioritised plan with owners and timing, stage, a blended win probability and a champion test. Notes added later re-score the deal and a history shows what moved. Pipeline insights name the deals to work this week, what is slipping, stale or missing from the forecast; the Monte Carlo forecast turns the probabilities into a revenue band. On the page, in chat and over MCP.

    Shipped 16 Oct 2026Open
  • Deal assessment and house-style deal reportsNexus

    Fixed rules read every deal's scores like a pipeline review. A verdict, a stage gate that says whether the stage is earned and what the next one needs, red flags for the patterns that lose deals, and the three levers that add the most weighted score. The deal report downloads as Word or PDF in the Enaible U house style. On the page, in chat and over MCP.

    Shipped 18 Oct 2026Open
  • Nexus reports on every runNexus

    Every cascade run files a rendered report with the full step output, grouped per persona, and idle runs stay quiet so the report list only holds work.

    Shipped 31 Mar 2026Open
  • Mission ControlNexus

    One view over every cascade, task and agent with step drill-down, schedule pills and live job events in the notification bell.

    Shipped 31 Mar 2026Open
  • AI workforce rosterNexus

    A living, seeded roster of the agents with their inner thoughts, a chat per agent, and skills an agent can request and an operator can grant.

    Shipped 26 Aug 2026Open
  • Work reviewNexus

    Agent output lands in a review queue where an operator approves it, awards bonus XP and sees the verifier's grade before anything is taken as fact.

    Shipped 31 Mar 2026Open
  • Nexus rewardsNexus

    Every agent earns XP per completed task with an event log of when and why, levels up as it accumulates, and can be granted bonus XP from the workforce view.

    Shipped 29 Mar 2026Open
  • Librarian research engineNexus

    Autonomous background research with a grounding gate. Every report is verified against the evidence it gathered, graded A, B or C, has dead links flagged, and a company's own site never counts as independent confirmation.

    Shipped 16 Sept 2026
  • Topic classification before researchNexus

    The Librarian classifies a topic before it starts and uses a brief and an acceptance bar specific to that kind of topic.

    Shipped 16 Sept 2026
  • Competitive intelligenceNexus

    Head-to-head comparisons and trained-knowledge dossiers on routed lanes, rendered as reports.

    Shipped 9 May 2026Open
  • Documents in the house style and your themePlatform

    Every Word, PDF and printable report is drawn in one house style and in the colours of the theme selected in the app, darkened where print needs it, with markdown reports rendered into the same style.

    Shipped 19 Oct 2026
  • Self-maintaining documentationNexus

    The Documenter covers all ten lanes of the documentation site, refreshes every page's screenshot nightly, derives prose, controls and API endpoints from the source, rewrites only the sections that drifted, leaves moved routes alone, and a test fails when a documented endpoint or control stops existing or a lane has no nightly run.

    Shipped 16 Sept 2026Open
  • Job schedulerNexus

    Cron-style scheduling for time-based and event-triggered jobs, with a run-on-startup flag and a failure model that never disables a job permanently after one bad run.

    Shipped 14 Mar 2026Open
  • Operations pageNexus

    The numbers that used to need SSH: what fails quietly, what is stuck and what is falling behind, in one page.

    Shipped 23 Sept 2026Open
  • Kanban todo boardSentinel

    Intake, in progress, done and archived columns with drag and drop, nested sub-todos, tags, priority and due dates. Agents move cards through MCP.

    Shipped 25 Mar 2026Open

Memory and context

Persistent, user-scoped memory that gets more useful over time, plus the document and key-value stores agents read and write.

  • Persistent memoryNexus

    Memories survive sessions, are scoped to the user, deduplicated by content hash and enriched in the background with keywords, so the assistant remembers preferences, project state and history.

    Shipped 21 Mar 2026Open
  • Memory categories and duplicate mergeNexus

    A local model assigns categories and a nightly pass merges duplicates, so the memory store stays tidy without anyone curating it.

    Shipped 25 Aug 2026
  • Semantic document recallNexussovereign

    Documents are embedded locally and recalled by meaning as well as by keyword, with a disk cache so re-indexing is cheap.

    Shipped 18 Aug 2026
  • Memory graphNexus

    A visual graph of memories and the entities they mention, for seeing how what the assistant knows hangs together.

    Shipped 9 Apr 2026Open
  • Document graphNexus

    An intelligent file indexer with cross-references, headings, keywords and code blocks per document, searchable by agents through MCP tools.

    Shipped 19 Mar 2026Open
  • Key-value context storeNexus

    A local-first key-value store with dot-notation keys and TTL, readable and writable by agents through MCP tools and curated by a Context Curator cascade.

    Shipped 25 Mar 2026Open
  • Context budgetPlatform

    Every prompt is assembled within a token budget and the MCP bridge is a sensitivity boundary, so a model only ever sees what the operation needs.

    Shipped 18 Aug 2026

Local models and routing

Every operation runs on a named local model with a fallback, is traced, measured and accounted for, and never leaves the machine unless you say so.

  • Operation routing lanesPlatformsovereign

    Each operation names a primary and a fallback local model with thinking under routing control. Untagged calls fall back to the chat lane, and a local-only box is a configuration, not a failure.

    Shipped 18 Aug 2026Open
  • Ollama providerPlatformsovereign

    Routing can name local models served by Ollama, which waits for the GPU at boot and keeps two models resident so nothing is ever loaded partly on CPU.

    Shipped 18 Aug 2026
  • Model benchPlatformsovereign

    A fleet yardstick that records every ask, response, time and verdict, runs as a daily cascade and produces an Assessor brief, so promotions are decided on measurements.

    Shipped 19 Aug 2026
  • Interaction tracesNexus

    One linked trace per chat turn, cascade, tool execution and background job, with span trees, a waterfall, request and response capture and the operation each span served.

    Shipped 18 Aug 2026Open
  • Usage accountingNexus

    Per-operation usage from the traces, a daily trend, a model by operation matrix and a fleet-wide view where every caller, including standalone binaries, is tagged.

    Shipped 16 Sept 2026Open
  • OpenTelemetry loggingNexus

    A pluggable cascade step that writes OTEL logs to disk or a collector with standard resource attributes, and a log viewer page.

    Shipped 25 Aug 2026Open
  • Local image generationNexussovereign

    Image generation runs on the local GPU and is switched on from the dashboard, closing the last hosted door.

    Shipped 18 Aug 2026Open
  • Local PII detectionPlatformsovereign

    Identifiers are found locally by rules first and the remainder is routed to a local model, so personal data is classified without leaving the machine.

    Shipped 18 Aug 2026Open
  • Bring your own keySentinel

    A user can set their own provider key and default model and it takes precedence over the platform defaults, with requests, tokens and estimated cost tracked per user.

    Shipped 28 Mar 2026Open
  • AI Model Hub discovery ledgerSentinel

    Nightly model discovery records every run in a ledger shown on the page, flags fleet candidates for the Steward instead of pulling them, and stops entirely under the zero-egress switch.

    Shipped 27 Sept 2026Open

Sovereign security

Security built into the platform rather than bolted on, with an audit record of every action and secrets that never leave the server in the clear.

  • Immutable audit logPlatform

    Every action, including logins, tool calls, memory writes, vault reads and admin changes, is written to an append-only record with method, path, IP, user agent, status and duration.

    Shipped 29 Mar 2026Open
  • Session managementPlatform

    Every login is a tracked session with a device hint, IP address and last-seen time; all active sessions are listed, any of them can be revoked with one click, and a revoked session ends on the server at once.

    Shipped 29 Mar 2026Open
  • Two-factor sign-inPlatform

    Time-based one-time codes as a second factor on sign-in.

    Shipped 29 Mar 2026
  • Google and Microsoft sign-inPlatform

    Single-tenant Microsoft Entra ID and Google sign-in, with the server saying at boot which buttons are live.

    Shipped 17 Aug 2026
  • Sentinel vaultSentinel

    User secrets encrypted at rest with AES-256-GCM under a key derived from the passphrase with scrypt and a per-account salt, sealed automatically after inactivity and unsealed explicitly.

    Shipped 19 Mar 2026Open
  • Server-held vault devicesSentinelsovereign

    A remembered device holds only a random token in a cookie the page cannot read; the server keeps the vault key wrapped under the platform key beside the device name and expiry the profile lists and revokes.

    Shipped 15 Oct 2026
  • Page access per rolePlatform

    A matrix of dashboard pages against the account roles; an unticked page leaves the menu and answers with a notice when opened by address, for administrators by their role too. Visibility only; API guards are unchanged.

    Shipped 16 Oct 2026Open
  • AI vaultSentinel

    Secrets an agent may use, published from the personal vault one at a time, with tool-level restrictions, sensitivity flags and an access trail, exposed as MCP read, write and list tools.

    Shipped 19 Mar 2026Open
  • Complete account deletionPlatform

    An administrator deletes an account in one step: what it runs stops, its ways in and personal data go, and what the person did stays as records without an account.

    Shipped 13 Nov 2026Open
  • MCP on the OAuth 2.1 profilePlatform

    AI tools connect to the platform's MCP endpoint through the platform's own sign-in, with no outside identity provider: the person approves the tool on a consent page, the tool gets ten-minute tokens for that one service, optionally bound to its own key, and every use is recorded. The person sees and disconnects connected tools on Account, Security, in the chat and over MCP.

    Shipped 14 Nov 2026Open
  • Weather anywhere, with its sourcesPlatform

    Current weather and a forecast for any place in the world from Open-Meteo, with the national weather service behind it named and linked as numbered sources; in the chat, over MCP and through the API.

    Shipped 14 Nov 2026
  • World timePlatform

    The time anywhere in the world, from the server's own NTP-synced clock and the IANA time zone database: 24-hour time, zone, UTC offset, daylight saving and the next clock change, and times compared between places; in the chat, over MCP and through the API.

    Shipped 14 Nov 2026
  • Sign-in on the OAuth security baselinePlatform

    Google and Microsoft sign-in hold to RFC 9700 and RFC 9207: the provider is bound to the flow, an issuer the provider names must be its own, each provider can return to its own redirect URI, and every use of a platform API key is recorded, with scoped keys living 30 days by default.

    Shipped 14 Nov 2026Open
  • Chat and MCP share every capabilityPlatform

    Whatever the AI chat can do is on MCP, and whatever MCP can do is in the chat, through one implementation and the same access checks; a tool that acts for a person needs that person, and one that acts as the platform needs an administrator. Every capability without a page says why.

    Shipped 14 Nov 2026Open
  • Capability ledgerPlatform

    Everything the platform's AI can do, and on which channels (the chat, MCP, agent profiles, the page where a person does it), read from the code itself, with the reason for every channel a capability is not on.

    Shipped 13 Nov 2026Open
  • Governed, sealed chat instructionsPlatform

    The instructions every chat turn runs under are a sealed template in the framework, small by default with the tool groups a message needs; a turn with personal context stays on the local models, and every sealed record names the instructions it ran under.

    Shipped 13 Nov 2026
  • Memory scopes and relevant recallPlatform

    Memories are your own, the platform's, or a customer's, shared with everyone who works for that customer and nobody else. Each chat turn gets only the memories that mean something to the message, at most three, chosen locally by meaning; memories are extracted on the local models only.

    Shipped 13 Nov 2026Open
  • Memory privacy scopingPlatform

    Memories live under user-scoped paths so a cross-user read is impossible at the filesystem and sync layers, not just in the API.

    Shipped 21 Mar 2026
  • Scoped API keysPlatform

    API keys carry an identity and scopes, are rate limited per key, and are logged only as a short prefix.

    Shipped 21 Aug 2026Open
  • Firewall enforcement and the autonomous ban setPlatformsovereign

    The platform's bans reach the server's own firewall, capped at 20 new automatic bans a minute and governed by administrators only, with fail2ban's SSH bans shared, range escalation, stored strikes, behaviour signals, public blocklists and expiring automatic bans, each proven in dry-run first. Its own page previews what the firewalls would receive and what the safety checks refuse, before any firewall is switched on.

    Shipped 15 Nov 2026Open
  • Intrusion strikes and temporary bansPlatform

    Signature detections need three strikes, bans escalate and expire, known users are exempt, and a proxy address is trusted only when declared.

    Shipped 21 Sept 2026Open
  • Admin user managementPlatform

    Manage users and admin access from the dashboard with named roles that can only ever grant, and every change reaching the audit log.

    Shipped 24 Sept 2026Open
  • Daily vulnerability sweepOperations

    A daily cascade asks whether anything known to be vulnerable affects what actually runs, across the Go modules and npm packages, and files a task per finding.

    Shipped 21 Aug 2026
  • Published security.txt and llms.txtPlatform

    The platform publishes its security contact and its guidance for language-model crawlers at the standard locations.

    Shipped 25 Aug 2026
  • Personal data export and deletionPlatform

    A user can export or delete their own data from the account pages.

    Shipped 29 Mar 2026Open

Spectre site intelligence

A real, on-premise crawler with structured extraction, change watching, research and evidence-grade archiving, callable from the UI, the API and MCP.

  • Structured page extractionSpectre

    Any page becomes clean JSON and Markdown with title, headings, body text, internal and external links, word count and depth, with readability extraction and headless Chrome escalation when a page needs it.

    Shipped 31 Jan 2026Open
  • Domain crawl and site mapSpectre

    Crawl a page, a section or a whole domain within a depth and page budget, build the site tree as pages are found, and map a site from its sitemap.

    Shipped 31 Jan 2026Open
  • Schema extractionSpectre

    Extract fields from a page against a saved JSON schema or a prompt, on the spectre routing lane so usage accounting and routing edits apply.

    Shipped 28 Mar 2026Open
  • Change watchingSpectre

    Watch a page on a schedule, pause and resume it, and see a line diff when the content changes.

    Shipped 28 Mar 2026Open
  • Batch jobsSpectre

    Run many URLs as one asynchronous batch with bounded concurrency and a completion webhook.

    Shipped 28 Mar 2026Open
  • Multi-engine searchSpectre

    Web search across several engines with an answer synthesised on the spectre lane; hosted engines are used only when their keys are set.

    Shipped 28 Mar 2026Open
  • Image and keyword extractionSpectre

    Pull the images and the ranked keywords out of any page.

    Shipped 28 Mar 2026Open
  • Document conversionSpectre

    Convert PDFs and other documents to text and Markdown, singly or in batch.

    Shipped 28 Mar 2026Open
  • CVE feedSpectre

    Vulnerability feeds and a dependency scan, so Spectre can say whether a finding touches something you run.

    Shipped 6 Apr 2026Open
  • Hopper and seed listsSpectre

    Domains discovered during a crawl land in the hopper for labelling and chaining; seed lists start recurring crawls from a curated set.

    Shipped 25 Mar 2026Open
  • Web answers from the pages readSpectre

    Ask the chat something on the web and it answers from the few best pages Spectre opened for it, citing each as [1], [2] with a link; only pages that opened are sources, dead links are left out and named, and duplicates are shown once.

    Shipped 13 Nov 2026
  • Spectre ScoutSpectre

    Topic-driven automatic research with daily briefs, editable topics and themed report downloads. The report cites every claim to a numbered source, a claim no source carries is struck before any model judges it, its confidence comes from how many independent sites agree, and a verifier model checks it.

    Shipped 21 Aug 2026Open
  • Spectre phase 1, trustworthySpectre

    The fixes from the Spectre review. Spectre returns the page that was asked for, with its real status and user agent. Every read has a size limit. Watches run in a bounded pool and back off when they fail. Only a job's owner or an administrator can stop it. Search and zero egress hold per tenant. Every channel applies the same authorization.

    Shipped 30 Oct 2026Open
  • Spectre observability dashboardSpectre

    Spectre counts what it does where it happens. That covers fetches, cache hits, browser use, refusals by robots.txt, egress, SSRF and size, browser waits, watch runs, webhook deliveries and proxy fallbacks, with timings and per-host stats. The overview page shows them with a 24-hour timeline and the failing, slow and busiest hosts. The same snapshot reaches chat and MCP, and a Prometheus endpoint serves it to your own monitoring.

    Shipped 30 Oct 2026Open
  • Spectre joined upSpectre

    One runtime for every way Spectre reads the web. There is one link walk for Crawl, Map, the crawl extract and the harvester, all through the normal fetch. There is one browser runtime, with a fresh browser context per task and every request a page makes checked. There is one cache, bounded in memory and on disk. Guards added to the fetch now hold on every path.

    Shipped 31 Oct 2026Open
  • Spectre phase 2, reliable at scaleSpectre

    Spectre takes turns per site and honours crawl-delay, 429 and Retry-After. It follows robots.txt to RFC 9309 and a page's own noindex, nofollow and noarchive, and names itself in its user agent. Evidence is signed with a published Ed25519 key, chained from day to day, optionally timestamped and pruned after its retention. Pages are read in their own character set, sitemaps come from robots.txt and gzip, and watches use conditional requests. A report is never verified by the model that wrote it. Webhooks go through a durable outbox, and a test holds every capability to its four channels.

    Shipped 31 Oct 2026Open
  • Tender and regulation radarSpectre

    Standing questions to TenderNed by words and CPV codes, and to the Officiële Bekendmakingen by gazette, authority and words. The radar reports new tenders, rectifications and closing dates, new publications and new versions, and, for administrators, which fields of a watched KVK registration changed, asked through Bastion. Each run reads only what is new, under robots.txt and each source's pace, with the answer kept as evidence if asked. New tenders can go straight into the Tenders pipeline, and every theme has webhooks and is on all four channels.

    Shipped 31 Oct 2026Open
  • Search without egressSpectre

    Under zero egress search answers from the tenant's private index of the pages Spectre has fetched, ranked with keywords and the local embedding model on our own server: no search engine is asked and the query never leaves the platform. On the search page, in the scout, the AI chat and MCP.

    Shipped 8 Nov 2026Open
  • Competitive intelligence watchesSpectre

    Competitors' prices, features and limits followed on a board, read by CSS selector or by the local model on a schedule. Only material changes count (a price move of half a percent or more, a feature that reads differently), a value that could not be read never does, and a weekly digest and a price and feature table show what moved. On all four channels.

    Shipped 8 Nov 2026Open
  • Authenticated intranet crawlingSpectre

    Spectre reads and crawls a tenant's own intranet behind its login, through a read-only service account whose session an administrator brings. The session is sealed with the platform key, goes to its own host only, never through a proxy, and is never in an answer, log, evidence or cache key. Sign-in screens are recognised, pages join only the reader's private index and may be kept as signed evidence naming the source, and every use and refusal is recorded in Bastion. On the screens and API, in the AI chat and over MCP for a person; refused to an agent on its own by design.

    Shipped 12 Nov 2026Open
  • Legal-grade evidence exportSpectresovereign

    Spectre evidence as one WACZ package for a dispute, an audit or a supervisor: the archived files as the hash chain proves them, a replay copy for WARC viewers, the day chain with its signed root and timestamp, the public key and a statement in Dutch of what was captured when and how to verify it. Every file hashed and the list signed; a record that fails is packaged with what failed. On all four channels.

    Shipped 2 Nov 2026Open
  • Website compliance monitorSpectresovereign

    For government websites: each page visited in a fresh browser without accepting anything, checked for trackers and cookies before consent, what WCAG 2.1 AA lets a machine check and the accessibility statement, security headers, HTTPS and security.txt, and the Woo page with its information categories. Every finding names its rule, every check is scored, and every page is kept as signed evidence. Weekly by default, and on all four channels.

    Shipped 2 Nov 2026Open
  • robots.txt everywhereSpectresovereign

    One shared robots cache is consulted by fetch, crawl, extract, map, watch, batch and the reading of Scout's sources, crawl-delay is honoured, and the only override is an admin action that is recorded. Since 1.132.0 the link-health scan obeys it too. The scout's browser search still reads search-engine result pages without asking robots.txt; PROP-20261112-01 proposes to change that. Under zero egress search answers from the private index (since 1.131.0).

    Shipped 27 Sept 2026
  • Zero-egress guarantee modeSpectresovereign

    A tenant policy that refuses every hosted lookup (hosted search, the Wayback Machine, Discord notifications) and stamps every result with an egress attestation. Under zero egress a page rendered in the browser may not load resources from other sites: the browser refuses them and the attestation lists what was refused. Personal data is always redacted under zero egress; a request cannot switch that off (since 1.133.0).

    Shipped 27 Sept 2026
  • PII-aware extraction with the vaultSpectresovereign

    The local PII detector runs over every page, and findings are replaced by placeholders before results reach a model, disk or a webhook when the tenant asks for it, always under zero egress.

    Shipped 27 Sept 2026
  • Evidence-grade archivingSpectresovereign

    Raw response, headers, rendered DOM and screenshot per fetch as WARC with the real status and user agent, hash-chained by day and from day to day, each day root signed with the platform Ed25519 key whose public half is published, an optional RFC 3161 timestamp per day, pruning after the retention with the chain kept, and a verify endpoint and MCP tools to fetch and verify evidence.

    Shipped 27 Sept 2026
  • Dutch government source packsSpectresovereign

    Packs for TenderNed, officielebekendmakingen.nl and the KVK register with seeds, a schema, a watch preset and legal notes, to read and apply by hand; running them on a schedule is the Tender and regulation radar (since 1.121.0).

    Shipped 27 Sept 2026
  • Intranet crawling under the hub's SSRF policySpectresovereign

    Spectre crawls SharePoint, Confluence and internal wikis on the same network under an allow-list with redirect and DNS-rebind protection.

    Shipped 27 Sept 2026
  • Route every Spectre inference through the spectre laneSpectre

    Schema extraction and search answers run on the spectre routing lane instead of the default chain, so routing edits and usage accounting apply to them.

    Shipped 27 Sept 2026
  • Persisted crawl policiesSpectre

    Batch and watch policies are stored with the job, so a run keeps the conduct it was started with.

    Shipped 27 Sept 2026
  • Format selector on FetchSpectre

    Ask Fetch for markdown, html, raw html, links, a screenshot or json and get only that back from the cached capture, so a caller never pays for formats it does not read. Screenshots are reachable from the API.

    Shipped 27 Sept 2026Open
  • Declarative page actionsSpectre

    Wait, click, type, press, scroll, run a script, take a screenshot and scrape as named steps in the same Chrome session, each recorded in the result, so login walls and infinite scroll are crawlable without a raw script.

    Shipped 27 Sept 2026Open
  • Crawl-wide schema extractSpectre

    One schema runs over every page of a crawl or a URL list as a job, chunked and merged into one result and validated with a retry, on the local spectre lane.

    Shipped 27 Sept 2026Open
  • Per-request crawl controlsSpectre

    Limit, depth, include and exclude paths, subdomains, external links, sitemap mode, query-parameter handling and delay are set on each crawl, map and extraction instead of once for the whole tenant, and concurrency on each crawl-wide extraction.

    Shipped 27 Sept 2026Open
  • Caller-controlled cachingSpectre

    Each request says how old a cached capture may be, how fresh it must be and whether the new capture is stored, and every answer carries the time it was captured.

    Shipped 27 Sept 2026
  • On-demand change trackingSpectre

    A change-tracking format on Fetch compares a page with its previous capture and answers with the previous capture time, a status, a git-style diff and a schema-driven field diff, with tags for separate histories.

    Shipped 27 Sept 2026Open
  • Semantic change summaries in WatchSpectresovereign

    After a hash change a local model says in three sentences what changed and gives a materiality score, so cosmetic diffs stay below a threshold and the real ones get noticed.

    Shipped 27 Sept 2026Open
  • Signed lifecycle webhooksSpectre

    Started, page, completed and failed events for batches, extractions and watches are delivered with an HMAC signature header and retried, so a receiving system can trust and replay them.

    Shipped 27 Sept 2026
  • Job cancel, errors and paginationSpectre

    Every check, tracking fetch, batch and extraction is a job that can be cancelled, lists its errors and pages its results with a cursor.

    Shipped 27 Sept 2026Open
  • Batch improvementsSpectre

    Batches drop duplicate URLs, take their own formats and schema, respect the browser pool when running concurrently and page their results.

    Shipped 27 Sept 2026Open
  • PDF and Office auto-parseSpectresovereign

    A link to a PDF, Word, Excel or CSV file is read as a document with page markers and a parser-and-confidence record, and a scanned PDF is transcribed on the local vision lane.

    Shipped 27 Sept 2026Open
  • Browser tiers and Chrome proxyingSpectre

    Fetches run under a basic, full or auto browser tier, auto retrying with the longer full render when a page did not render, and each tenant's proxy pool is passed into Chrome. Since 1.136.0 no tier disguises the browser: Spectre always names itself, and a page that refuses the crawler is reported as protected, never retried in disguise (PROP-20261112-01).

    Shipped 27 Sept 2026Open
  • Location and language emulationSpectre

    A country and a language list set the Accept-Language header, the Chrome locale and timezone and pick a matching proxy, so a page is seen as a local visitor sees it.

    Shipped 27 Sept 2026
  • Vision-assisted extractionSpectresovereign

    Charts are captioned, scanned notices transcribed and cookie-walled pages checked by the local vision model, and the captions feed back into image extraction as alt text.

    Shipped 27 Sept 2026Open
  • A real MCP server for SpectreSpectre

    Scrape, crawl, map, extract, the semantic and private search, Scout, watch changes, evidence, the radar, compliance, competitive and intranet tools are served over MCP with streamable HTTP and a stdio proxy for Claude Code and Cursor, behind the same scoped API keys, so outside agent runtimes call Spectre natively. Creating a watch and the multi-engine web search are not on MCP yet.

    Shipped 27 Sept 2026Open
  • Public API and SDKsSpectre

    A versioned Spectre API with scoped, quota-metered keys and per-key limits, an OpenAPI description and thin Go, TypeScript and Python clients.

    Shipped 27 Sept 2026Open
  • Semantic keywords and site searchSpectresovereign

    Every fetched page is embedded on local models that never leave the server, keywords are ranked by meaning instead of frequency, and everything a tenant has crawled is searchable by that tenant, with near-duplicate detection.

    Shipped 27 Sept 2026Open
  • Spectre MCP toolsSpectre

    Crawl, extract, map, Scout, images, link health, evidence, radar, compliance and more are callable by agents as MCP tools, and by API key from outside; the platform's own tools (CVE, Discord, telemetry, the harvester's intelligence) stay inside. Document conversion is on the screens and the API.

    Shipped 6 Feb 2026Open
  • Discord toolsSpectre

    Agents can post to a Discord channel or send a direct message through the platform's bot.

    Shipped 14 Mar 2026

The AI-owned mailbox

A mail service run by AI, with its own SMTP server, an autonomous responder that starts in draft mode, and a decision record for every stage.

  • Own SMTP serverMail

    The platform receives mail on its own listener with IP blacklist, honeypots and relay protection, and sends through a direct relay.

    Shipped 25 Jan 2026Open
  • Keep the raw message and real headersMail

    The raw message is kept as a file with the full header map and attachment list, and bodies are no longer truncated.

    Shipped 27 Sept 2026
  • ThreadingMail

    Message-ID, In-Reply-To and References are stored and emitted, a thread id is computed, and threads show in the mail page.

    Shipped 27 Sept 2026
  • Stop banning legitimate sendersMail

    An unknown recipient gets a plain refusal instead of a blacklist strike, so a legitimate mail server delivering to the AI address is no longer treated as a relay probe.

    Shipped 27 Sept 2026
  • Autonomous responder with a held queueMailsovereign

    Guard, understand, draft, verify on a different model family, decide, send and record. Every mailbox starts in draft mode, so replies wait in a queue with Approve, Edit and Reject until an operator chooses autonomy per mailbox.

    Shipped 27 Sept 2026Open
  • Loop protectionMail

    Auto-submitted, bulk, list, bounce and self-addressed mail is never answered, and AI replies are marked so other systems do not answer them either.

    Shipped 27 Sept 2026
  • Allow, deny and suppression listsMail

    Sender and domain policy per mailbox plus a suppression list, checked before any draft is written.

    Shipped 27 Sept 2026
  • Decision audit trailMailsovereign

    One record per stage with the model, lane and prompt hash, shown as a timeline on the mail and counted in the usage tables.

    Shipped 27 Sept 2026
  • Deliverability cardMail

    The mail page shows the state of the domain's MX, SPF, DKIM and DMARC records, so a missing record is visible before anyone wonders why nothing arrives.

    Shipped 27 Sept 2026
  • Generate and wire DKIM keys in deployMail

    When a selector is set the deploy creates the DKIM key, wires it in and prints the TXT record to publish.

    Shipped 27 Sept 2026
  • Kill switch, retention, one send pathMail

    Every way mail leaves goes through one send path with the outbound switch, the workforce kill switch, the sender check and the caps. After the retention term an operator sets (MAIL_RETENTION_DAYS), a mail's content and raw message are removed while the record that it was there stays; without a term nothing is removed.

    Shipped 12 Nov 2026Open
  • Correspondent roleMail

    The framework binds the mailbox to a Correspondent role with answer-only authority, hold by default, quotas and an escalation path.

    Shipped 27 Sept 2026
  • PII-aware storageMailsovereign

    Mail is scanned for personal data at ingest, redacted before any model call, and stored bodies are encrypted at rest.

    Shipped 28 Sept 2026
  • Rate limits and daily capsMail

    Every mailbox has an hourly and a daily cap in recipients, counted the same whether the mail comes from the inbox, the chat tool, an agent, a cascade or a mail client through the relay.

    Shipped 30 Oct 2026Open

Robeeflow integration hub

A Dutch-government-grade integration hub for Digikoppeling, Peppol and partner exchange, with flows, connectors, certificates and evidence, all in your own PocketBase.

  • Robeeflow inside the platformRobeeflow

    The hub runs as a self-contained plugin with its own PocketBase in the platform's storage tree, proxied under the platform origin and gated to admins, and bridged into the AI, logging and reporting.

    Shipped 28 Aug 2026Open
  • Standard ExplorerRobeeflowsovereign

    One standard as a learning center, everything sourced. Its purpose explained from the stored sources, every message type with its structure, every WSDL or REST operation as request, response and fault, namespaces and endpoints, clickable samples, versions with a computed comparison, and a guided first walk for newcomers.

    Shipped 15 Nov 2026Open
  • Standards Studio loopRobeeflowsovereign

    The Standards Guardian files and starts the gap assignments on its own, rung by rung up to a mapping onto the canonical model and a dry-run sample flow, local models review them, and only a person activates each rung; pausable at any time.

    Shipped 14 Nov 2026Open
  • Standards StudioRobeeflowsovereign

    A catalogue of whole integration standards with coverage per message, assignments built from step pills, a four-step review where only a person activates, and a live view of the agents handing work to each other.

    Shipped 14 Nov 2026Open
  • Flow designerRobeeflow

    A visual designer with the full node palette, including Validate, Transform, HTTP Request, Log and Schedule nodes, so time-triggered flows run without code.

    Shipped 26 Sept 2026Open
  • Flow runsRobeeflow

    One row per execution with a node timeline and a re-run button.

    Shipped 26 Sept 2026Open
  • Flow versions with rollback, dry runs and error edgesRobeeflow

    Every save is a version you can roll back to, a flow can be dry-run before it goes live, and an error edge routes a failed node instead of aborting the run.

    Shipped 27 Sept 2026
  • Configuration as codeRobeeflow

    Export the hub's configuration as a bundle with secrets removed, and import one with a diff before anything changes.

    Shipped 27 Sept 2026
  • Digikoppeling ebMS 2.0Robeeflow

    Outbound ebMS 2.0 with reliable delivery, an inbound listener over two-way TLS, a delivery screen and a send node.

    Shipped 25 Sept 2026Open
  • Digikoppeling WUSRobeeflow

    Signed synchronous SOAP 1.2 in both directions.

    Shipped 27 Sept 2026
  • Digikoppeling REST-API profileRobeeflow

    OAuth 2.0 with private_key_jwt, mTLS entry points and REST entry points that run flows.

    Shipped 26 Sept 2026
  • AS4 and PeppolRobeeflow

    An AS4 / ebMS 3.0 profile puts Digikoppeling ebMS3 and Peppol on the same outbox, inbox and trust store, with SML/SMP dynamic discovery at send time and encrypted payloads.

    Shipped 26 Sept 2026
  • Digikoppeling Grote BerichtenRobeeflow

    Large payloads travel out of band under the Grote Berichten profile.

    Shipped 26 Sept 2026Open
  • WS-Security signingRobeeflow

    Messages are signed and verified with WS-Security on the request path.

    Shipped 28 Aug 2026
  • OIN registerRobeeflow

    A daily mirror of the Logius OIN register with a public API and a lookup on the agreements and partners pages.

    Shipped 24 Aug 2026Open
  • CPA importRobeeflow

    Import an agreement from a cpa.xml file instead of typing it.

    Shipped 26 Sept 2026Open
  • Certificate lifecycleRobeeflow

    An inventory of everything that expires, alerts, CSR generation, install and rotation, with signing certificates verified against PKIoverheid and revocation checked through OCSP with a CRL fallback.

    Shipped 26 Sept 2026Open
  • Trust anchorsRobeeflow

    A screen for the trust anchors with checks at entry.

    Shipped 25 Sept 2026Open
  • Message trackingRobeeflow

    Volumes, partners, acknowledgement latency, what is stuck and a per-message trace.

    Shipped 26 Sept 2026Open
  • Message archive and retentionRobeeflow

    Received and sent messages are archived and expire under a retention policy.

    Shipped 26 Sept 2026Open
  • Message replay and backpressureRobeeflow

    Replay a received message and hold an endpoint back when it is overwhelmed.

    Shipped 26 Sept 2026Open
  • Masked replay sandboxRobeeflowsovereign

    Replay a real received message into a dry run with personal data masked, so an integrator can debug production traffic without seeing it.

    Shipped 27 Sept 2026Open
  • Delivery alertsRobeeflow

    Dead letters and rejections raise an announcement and a mail, once per agreement.

    Shipped 27 Sept 2026
  • Hub secrets vaultRobeeflow

    Keys, API keys, tokens, partner credentials and connector licences are sealed at rest with a lookup hash beside them and masked on read.

    Shipped 26 Sept 2026
  • Message bodies sealed at restRobeeflow

    Stored message bodies are encrypted, and the vault degrades to a visible warning rather than taking the hub down when the key is missing.

    Shipped 27 Sept 2026
  • Cloud connectorsRobeeflow

    An outbound-only on-premise connector runner enrolled from a screen, with a work channel, a job queue with retry and cancel, and signed requests that are accepted once.

    Shipped 25 Sept 2026Open
  • Connector sourcesRobeeflow

    A connector picks up files from a customer's SFTP server, a folder or an HTTP endpoint, provisioned from the hub, with host keys pinned and source secrets sealed.

    Shipped 27 Sept 2026
  • Counterparty test harnessRobeeflow

    A loopback conformance run and a simulated partner for connection tests.

    Shipped 26 Sept 2026Open
  • OpenAPI importRobeeflow

    Import exit points from an OpenAPI description.

    Shipped 26 Sept 2026Open
  • AS4 message service handler checked against the standardRobeeflow

    ebMS 3.0 AS4 on the hub, measured per agreement against the OASIS AS4 profile, EU eDelivery AS4 1.15 and 2.0, Peppol AS4 2.0 and Digikoppeling identities on ebMS3. Push and pull, receipts with non-repudiation in the response or on a callback, reception awareness with duplicate detection, gzip, several payloads, the test service, signatures with RSA, ECDSA or Ed25519 that must cover the header, body and every part, and AES-128-GCM payload encryption under RSA-OAEP or X25519 key agreement with HKDF and key wrap. One page shows the P-Mode, the findings with their clauses, the capability statement and the error table; a profile is applied to an agreement with one click.

    Shipped 17 Oct 2026Open
  • Go-live readiness pageRobeeflow

    One page that says whether the hub is ready for production, including interoperability and trust anchors.

    Shipped 27 Sept 2026Open
  • Health check that can say noRobeeflow

    The hub's health endpoint reports degraded when something it depends on is missing, instead of always answering fine.

    Shipped 25 Sept 2026
  • Safe to run more than one hubRobeeflow

    Leased singleton roles, a claim-before-run inbox and a flow concurrency ceiling, so a second hub is safe.

    Shipped 27 Sept 2026
  • Data catalogueRobeeflow

    Every message definition's fields, types and required flags in one searchable view, derived from what already exists.

    Shipped 31 Aug 2026Open
  • Service levelsRobeeflow

    Response time and availability commitments per customer, so the logs have something to be judged against.

    Shipped 28 Aug 2026Open
  • Mapping editorRobeeflow

    Edit field mappings, inspect a connector and read a record's history in the hub.

    Shipped 30 Aug 2026Open
  • Agents drive RobeeflowRobeeflow

    The platform's agents operate the hub through MCP tools.

    Shipped 28 Aug 2026
  • Managed file transferRobeeflow

    Partner profiles with a pinned host key and sealed credentials, transfers over secure file transfer in both directions, over HTTPS and from watched folders, encryption at the payload level, checksums that travel with the file, retries with backoff and a restart from the byte a transfer stopped at.

    Shipped 28 Sept 2026Open

Bastion

Personal data of people, buildings and companies lives in vaults behind one access layer on your own server, read under a registered purpose, protocolled on every access and reachable by AI only through a bridge that keeps the data local.

  • Bastion moduleBastion

    The personal-data vaults, the AI bridge and their console are one module of the dashboard beside Nexus, Sentinel and Spectre, with an overview of every vault's records, key state and rotation, the protocol chain check, the grants waiting for approval and the reads attested recently, and a page per part.

    Shipped 27 Sept 2026Open
  • BSN masked everywhere, never sent outBastionsovereign

    A BSN is stored masked with a keyed hash in every log, run and archive, is revealed only inside Bastion under a purpose, a reason and a second approver, and is masked before anything leaves the platform; only the integration that must deliver it to a government counterparty sends it.

    Shipped 14 Nov 2026
  • Vaults pageBastion

    One card per vault with its record count, purposes, protocol rows and key version; records are shown masked by field class and a field is revealed only with a recorded reason. The developer Vault Explorer is retired.

    Shipped 27 Sept 2026Open
  • Registered vault purposesBastion

    A purpose and its lawful basis are registered per vault and per consumer before any read, free text is refused, and a purpose that touches the BSN needs a Guardian's approval.

    Shipped 27 Sept 2026Open
  • Protocollering per readBastion

    Every read of the people vault writes a hash-chained protocol row with who asked, under which purpose, which fields came back and a hash of the record, without a raw identifier. The archiver never touches the chain and a person's protocol can be listed.

    Shipped 27 Sept 2026Open
  • Envelope encryption with per-vault data keys and rotationBastion

    Each vault has its own data key wrapped by a platform key, every record carries its key version, and rotation re-wraps the keys without re-encrypting every record.

    Shipped 27 Sept 2026
  • Keyed hashing and tokenisation serviceBastion

    A hashing key separate from the encryption key produces the lookup hashes for BSN, KVK and address keys, every BSN passes the elfproef first, and the hub's logs only ever show a hash prefix.

    Shipped 27 Sept 2026
  • Data-subject rights toolingBastion

    Look a person up under a registered purpose and produce an access export with protocol history, rectify with before and after recorded, reveal with a reason, and delete with a tombstone so a later import cannot bring the person back.

    Shipped 27 Sept 2026Open
  • AI chat bridge with vault tools for local modelsBastionsovereign

    Ask the chat about a person, building or company and the local model calls a server-side vault tool whose result is filtered and redacted by class before it enters the prompt, cited back to the record and its protocol row. An unlock ceremony precedes the first read and a hosted model lane is refused by code.

    Shipped 27 Sept 2026Open
  • MCP vault tools with scoped keysBastion

    External agents call vault search, get and verify over MCP with a key scoped to one vault, purpose, field set and rate, under the same checks and the same protocol and lineage rows as chat.

    Shipped 27 Sept 2026Open
  • AI-to-AI autonomous vault accessBastionsovereign

    An agent or flow step queries a vault for another agent under that agent's permission manifest, receives only the allowed fields, and both agents are linked to the record in the lineage. BSN-class fields never reach a step that sends outward.

    Shipped 27 Sept 2026Open
  • AI bridge keys stored as hashesBastionsovereign

    Every AI-side caller, whether a chat tool, an MCP key, an agent or a flow node, holds a scoped bridge key that exists on the server only as a keyed hash. A second approver is needed to unlock a BSN purpose.

    Shipped 27 Sept 2026Open
  • Data-never-leaves attestation per queryBastionsovereign

    Every vault read through chat, MCP or an agent returns a signed egress attestation naming the local lane and model, the fields and classes returned and the protocol and lineage identifiers, so an auditor can see the data never left the server.

    Shipped 27 Sept 2026
  • Field-level classification schema per vaultBastion

    Every field of every vault carries one of six classes, from public to secret, and that class decides what is masked on a screen, what a purpose may receive, what may reach a model and what may be exported. A field the schema does not name fails the read closed rather than passing as harmless.

    Shipped 28 Sept 2026Open
  • Purpose binding per vault with a generated verwerkingsregisterBastion

    The register of processing activities required by article 30 is rendered from the registered purposes, the classification schemas and the vault configuration alone, versioned by content hash, signed, and regenerated whenever a purpose or a retention term changes, so it cannot quietly drift from what the platform does.

    Shipped 28 Sept 2026Open
  • DPIA generatorBastion

    The data protection impact assessment is drafted from the same configuration, with the three mandatory triggers read from it rather than asked of anyone, the mitigations already in force listed with links to their evidence, and the open risks named. It stays a draft until a person signs it.

    Shipped 28 Sept 2026Open
  • Vault console with break-glass and no direct database accessBastion

    Revealing a citizen service number or a special-category field needs a second administrator to approve a request that is valid for fifteen minutes and for one reveal, and the request, the approval and the reveal are all recorded. Break-glass without a second person takes an explicit flag and raises an alert.

    Shipped 28 Sept 2026Open
  • Retention and archival per vaultBastion

    Every vault has a retention term, from the archive selection list or the customer's own decision, and a legal hold that freezes deletion until a recorded decision lifts it. A daily sweep respects both and the page shows what is past its term before anything is removed.

    Shipped 28 Sept 2026Open
  • Breach detection and alertingBastion

    Seven rules over the access record, from bulk reads and off-hours reveals to failed decrypts, break-glass and a broken chain, raise an alert and a notification, and the serious ones open an incident carrying the seventy-two hour clock from the moment awareness began. A rule counts rows and hashes and never names a person.

    Shipped 28 Sept 2026Open
  • Haal Centraal BRP connectorBastion

    Ask the population register live for one person or a search, with the fields taken from the registered purpose and the authorisation so nothing unauthorised is even requested, over the government transport with your own certificate. The answer is written into its vault and both the acquisition and the provision are recorded.

    Shipped 28 Sept 2026Open
  • BAG connectorBastion

    Address and building lookups from the national buildings and addresses register, kept within the published rate limits before the call rather than after a refusal, and cached only for freshness. A person or a valuation links to a building identifier instead of a free-text address.

    Shipped 28 Sept 2026Open
  • WOZ connectorBastion

    Property valuation queries limited to the fields the supervisory authorisation matrix gives your role, which narrows the answer once more after the purpose and its field list. Every value is stored with its reference date and the authority it was obtained under.

    Shipped 28 Sept 2026Open
  • KVK connectorBastion

    Company profiles from the Chamber of Commerce search, basic profile and establishment profile. The registration of a sole trader or a partnership is stored as personal data, because their trade name is often their own name and their registered address often their home, while the facts about the registration stay public.

    Shipped 28 Sept 2026Open
  • OIN register upgradeBastion

    The organisation identifier register is mirrored into a vault of its own through the access layer with a daily refresh, keeping parent and sub-identifier relations, the Chamber of Commerce cross-link and a freshness stamp and provenance on every lookup.

    Shipped 28 Sept 2026Open
  • Records belong to a customer and an authorisation covers a populationBastionsovereign

    A record names the customer it belongs to, and an authorisation names the people it covers rather than the whole register, the way a Dutch autorisatiebesluit does. A read is answered for the caller's own customer, an authorisation can be narrowed further to one municipality or any other field of the record, and reaching across customers is a declared, operator-only act that says so on every row it writes.

    Shipped 28 Sept 2026Open
  • Browsable record cards that carry no contentBastionsovereign

    A card per record showing its lookup hash, the key version that sealed it, when it was written and which customer it belongs to, and no field of the record at all, which is the most a list may lawfully see. The card carries the hash a read needs, so a record is opened without anybody having to know a citizen service number, and the number itself stays behind the reveal step. Each page of cards is one recorded read naming the page rather than a person.

    Shipped 29 Sept 2026Open
  • Assigning records to the customer they belong toBastion

    An operator gives an owner to records that have none, either one customer for everything matched or by a field of the record, so a municipality is given the people registered in it the way an autorisatiebesluit reads. The run counts before it changes anything, reports which values it could not place, and is reached from the console, the API, agent to agent, over MCP and in chat.

    Shipped 29 Sept 2026Open
  • Withholding a record from disclosureBastionsovereign

    A person can ask for their record to be withheld, which the population register calls a verstrekkingsbeperking. It then disappears from every read, search and count, and a single time-boxed look at it needs a second administrator to approve. A record that is out of scope, withheld or genuinely absent answers in exactly the same words, so the answer itself reveals nothing.

    Shipped 28 Sept 2026Open

The governed document store

A file about a person is personal data that happens to be a file, so it lives under the same law as the vaults. Hashed on arrival, sealed under its own key, carrying exactly one class, never overwritten and registered on every open.

  • Content-addressed document store with a signed daily rootDocument Storesovereign

    Every file is hashed on arrival and kept once under that hash, and every open, download, share, hold and deletion is one row in a per-day chain whose root is signed, so one verifier answers for documents, audit rows and vault reads alike.

    Shipped 28 Sept 2026Open
  • Per-document envelope encryption on the vault key hierarchyDocument Storesovereign

    Each document is sealed with its own data key, wrapped by its tenant key, wrapped in turn by the platform master key, so a rotation re-wraps keys without touching a single document and a sealed tenant cannot be read by the operator.

    Shipped 28 Sept 2026
  • Metadata schema aligned with MDTO and the six classesDocument Store

    Every document carries the archival elements a Dutch archive expects, from creator and retention term to checksum and file format, plus exactly one of the six classes that decides masking, sharing and export. A document without a class does not save, and the content scanner may raise that class but never lower it.

    Shipped 28 Sept 2026Open
  • Versions, immutability and tombstonesDocument Store

    A new upload of a known document becomes a new version linked to the previous one, and a deletion leaves a tombstone with actor and reason, so the same bytes cannot quietly come back and content disappears only once retention and holds allow it.

    Shipped 28 Sept 2026Open
  • Governed share links with recipients, expiry, limits, view-only and watermarkDocument Storesovereign

    Share a document with a named recipient under a purpose, an expiry, a download limit and an optional view-only mode with a watermark. The link is shown once and kept only as a hash, so a lost link is reissued rather than recovered, revocation is immediate, and every open and every refusal is recorded.

    Shipped 28 Sept 2026Open
  • One access layer for documents on all four channelsDocument Store

    Every list, read, upload, share and deletion passes one layer on your own server that resolves who is asking, applies the class rules, decrypts, writes the row and attests what left. The console and the API, the agent steps, the MCP tools and the chat tools are doors onto it and hold no rule of their own.

    Shipped 28 Sept 2026Open
  • Protocollering per open, download, share and deliveryDocument Storesovereign

    Every read on any channel writes a tamper-evident row with actor, purpose, channel, document hash and version, class and outcome, kept append-only for its legal term, so a steward answers who saw what and a person can list the rows about their own documents. A read whose row cannot be written is refused and no byte leaves.

    Shipped 28 Sept 2026Open
  • Retention per class and legal hold for documentsDocument Store

    Each document inherits a retention term from its archival metadata or its class, a daily run tombstones what is past term and removes the bytes after a grace period, and a legal hold for a dispute, a public-records request or an investigation freezes both until a recorded decision lifts it.

    Shipped 28 Sept 2026Open

Layer 7 gateway policies

Design, grade, simulate and publish API gateway policies, and sweep the endpoints they protect.

  • Policy designerLayer 7

    A node-graph designer with tree and canvas views, a palette built from the 202-assertion reference, editable author comments and a byte-identical serializer.

    Shipped 16 Aug 2026Open
  • Publish to a live gatewayLayer 7

    Publish a designed policy to a gateway with a preview first, and pull policies from a GitHub repository over a deploy key.

    Shipped 17 Aug 2026
  • Policy lint and security baselineLayer 7

    Every authored policy is graded against a security baseline and passes both checkers automatically before it can ship.

    Shipped 26 Aug 2026Open
  • Compliance gradeLayer 7

    Policies are graded against BIO/BIG, NIS2 and the NCSC web guidelines.

    Shipped 21 Aug 2026
  • Policy simulationLayer 7

    Run a policy against saved cases and a suite of them, and see which node decided the outcome.

    Shipped 16 Aug 2026Open
  • OWASP endpoint sweepLayer 7

    A weekly sweep of the app's own API surface and declared targets with scan levels L0 to L4, findings with history, lifecycle and muting, and an accepted-adjusted grade.

    Shipped 22 Aug 2026Open
  • FSC inway and outway policiesLayer 7

    Ready-made policies for an FSC inway and outway with token introspection, graded at A on the security baseline.

    Shipped 26 Aug 2026
  • Gateway health report per policyLayer 7

    The health check report carries an appendix per policy with its grade, the controls it fails, the improvement and tip for each, and the line inside the policy where the finding sits.

    Shipped 15 Oct 2026
  • Policy relationships and context scoringLayer 7

    A Relationships page draws how services, fragments and encapsulated assertions work together and which services reach each building block. Building blocks are scored through the services that use them, and portfolio security is averaged over the services. In the health check, in chat and over MCP as well.

    Shipped 19 Oct 2026Open
  • Health check in the house styleLayer 7

    The health check document opens with a management summary (key figures, grade spread, most common control gaps, the policies to fix first) and is laid out in the Enaible U house style, with coloured grades and severities and every automatic finding marked as such.

    Shipped 18 Oct 2026
  • Scheduled health check as a managed serviceLayer 7

    Per workspace, on a day of the month, the health check runs on its own, compares itself with the month before (health and compliance, new and resolved findings, certificates about to expire, changes on the gateway) and mails the Word report with a trend chapter and chart to the recipients a person set. Every month is kept with its report; chat and MCP read the runs and can take one without sending it.

    Shipped 19 Oct 2026Open
  • Health check figures and top tenLayer 7

    The health check ranks the ten services with the widest reach with their grades, draws how the services share the most used building blocks, and draws the routing map from the services to their backends with plain HTTP in red. Drawn on our own server in the app theme's colours; Mermaid diagrams in the Markdown version.

    Shipped 19 Oct 2026
  • Relationship and routing map per policyLayer 7

    Every policy in the health check appendix carries its own relationship map (who calls it, what it calls, which services reach it, calls to blocks missing from the export) and its own routing map (where its traffic goes, itself and through every building block it reaches, with plain HTTP in red). Figures in the Word report, the same in text in the Markdown version and for chat and MCP.

    Shipped 23 Oct 2026
  • One health check over all gateways of a customerLayer 7

    One Word or Markdown health check over two to six gateways of one customer, such as an external and an internal gateway. The gateways side by side, each with its own figures, and a chapter on what connects them. It shows routes from one gateway to the other, calls one gateway cannot resolve that the other can, and same-name policies that drifted apart. Appendices per gateway. From Layer 7 Settings, the API, chat and MCP.

    Shipped 23 Oct 2026Open
  • Report themes with embedded fontsPlatformsovereign

    Every app theme is also a Word document theme, with its typeface carried inside the file. That is Saira for every theme and Plus Jakarta Sans for Enaible U, regular, bold and italic, so a report looks the same on any computer. The themer is pluggable. Other projects and AIs list the themes, take a theme as a JSON pack with their own colours, fonts and wordmark, and render markdown into a themed Word document through the API with a key, chat, MCP or the stand-alone docxkit command line. On Admin, Design, Report Themes a person sees the themes side by side, downloads a sample and renders markdown.

    Shipped 24 Oct 2026Open
  • Variable intellisense in the policy viewerLayer 7

    Every ${variable} in the XML view and an assertion's properties says where it is set and read, flags a variable that is never set, read before it is set or never read, and jumps to each place; the variable findings in Policy health list each variable with its lines.

    Shipped 19 Oct 2026Open
  • Graphman bundle importLayer 7

    A Graphman, GMU or Restman bundle dropped on the workspace page becomes policies, services, encapsulated assertions, listeners and cluster properties.

    Shipped 15 Oct 2026Open
  • Resolved routing mapLayer 7

    Route targets written as variables and cluster properties are resolved through the bundle, alike hosts collapse into one node, and what cannot be resolved says so.

    Shipped 15 Oct 2026
  • Security analysis in policy contextLayer 7sovereign

    A policy is graded with what reaches it through Include fragments, encapsulated assertions, global policies and the gateway's listeners, so shared TLS and authentication count.

    Shipped 15 Oct 2026
  • Shared encapsulated-assertion rolloutLayer 7sovereign

    An audit or headers fragment written once as an encapsulated assertion and inserted at the top of chosen services, previewed as a graded diff and filed as a Restman bundle in the customer dossier. On the page, in chat and over MCP.

    Shipped 15 Oct 2026
  • AI capability suggestions on the local modelLayer 7sovereign

    The platform's own local model proposes the capabilities the fixed rules miss, each resting on an assertion in the policy and counting only once a person accepts it, beside a list of the assertions no rule recognises, custom and plugin assertions flagged.

    Shipped 30 Oct 2026Open

Federated Service Connectivity

A portable FSC node for the Dutch government's federated service standard, from management UI to interop runtime.

  • FSC managementFSC

    Identity, peers, directory, services and contracts managed from one module.

    Shipped 26 Aug 2026Open
  • FSC interop runtimeFSC

    Token authorisation, an inway and an outway, mTLS with certificate-bound tokens and a real inway listener.

    Shipped 26 Aug 2026
  • Contract signing and federationFSC

    RFC 8785 canonicalisation and detached JWS contract signatures, multi-peer federation, and a peer token endpoint with authorisation bound to the caller.

    Shipped 27 Aug 2026
  • Digikoppeling REST-API in FSCFSC

    The Digikoppeling REST-API 3.0.0 profile implemented on the FSC runtime.

    Shipped 27 Aug 2026
  • Token issuance the way the standard asksFSCsovereign

    Access tokens through the OAuth2 client-credentials form, carrying the grant hash and bound to the peer certificate, authorised against that grant on that contract, with the RFC 6749 error shape.

    Shipped 1 Oct 2026
  • Pass the official conformance suiteFSCsovereign

    A local test authority, a two-node federation on one machine and the Logius compliance tester run from a script, with a stored report per run.

    Shipped 1 Oct 2026
  • FSC Logging on the audit chain we already haveFSC

    The transaction log is written before a call goes through, an unlogged call is denied, the records sit on the platform's hash-chained audit files and a peer reads back its own.

    Shipped 1 Oct 2026
  • The peer-facing Manager APIFSCsovereign

    Every path FSC Core asks a Manager to serve to other peers over mutual TLS, now with the peer, peers and services listings, real cursor pagination and the standard's filters, and an operator view of what other peers see.

    Shipped 29 Oct 2026Open
  • Directory federation and peer discoveryFSCsovereign

    Read the group's FSC Directory, announce to it and publish services in it, and propose a connection to a listed service straight to the provider's Manager, with listed addresses never replacing the ones held.

    Shipped 29 Oct 2026Open

Addons, connectors and APIs

Everything is an addon with a manifest agents can read, and every route is documented.

  • Addon frameworkPlatform

    Addons declare their purpose, capabilities and governance in a manifest that the workforce reads, so an agent can find and use a module it has never seen.

    Shipped 17 Aug 2026
  • GitHub connectorPlatform

    A reusable connector over SSH deploy keys, wired into Layer 7 and cascades.

    Shipped 17 Aug 2026
  • Migration addonNexus

    Reads an Adeptia Connect export and produces a migration plan to Boomi: every flow with a diagram, every mapping classified by effort, a wave plan and a readiness score.

    Shipped 25 Sept 2026Open
  • AI policy reviewer with patch proposalsLayer 7

    Our own model proposes small, structured fixes for a policy's findings; each is verified by the same analysis, regression cases and standard requests before it is shown, and a person approves it into the next version.

    Shipped 19 Oct 2026Open
  • GitLab and Azure DevOps with pull requestsLayer 7

    Policies imported from and committed to GitHub, GitLab or Azure DevOps, proposed as a pull request carrying the change risk, health and security baseline, with the person who exported as the commit's author.

    Shipped 28 Oct 2026Open
  • Consumer API documentation from policiesLayer 7

    Documentation for the teams that call each service, generated from what its policy enforces: authentication, scopes, headers, the validated body, rate limits, the gateway's own responses and an example request, as Markdown to publish, with no secrets, IP ranges or backend addresses in it.

    Shipped 28 Oct 2026Open
  • OWASP API Top 10 coverage matrixLayer 7

    Every published service against the ten OWASP API risks, each covered, partial, gap or without evidence, from the policy controls in context, the scanner's findings and the API inventory, with what the gateway cannot show named per risk.

    Shipped 28 Oct 2026Open
  • Multi-gateway sync with drift alertsLayer 7

    Development, test, acceptance and production gateways per workspace, each checked on its own schedule against what was published or approved for it, with one notification when production stops holding what was approved and a comparison of the gateways name by name.

    Shipped 28 Oct 2026Open
  • API inventory and OpenAPI reconciliationLayer 7

    What a gateway publishes set against the customer's OpenAPI and Swagger descriptions and what the scanner reached: every service documented, shadow, zombie or disabled, the methods no description lists, and the documented operations no service exposes.

    Shipped 28 Oct 2026Open
  • Certificate and secret expiry radarLayer 7

    Every certificate a gateway workspace holds with its expiry, the services that depend on it and alerts at 90, 30 and 7 days; stored passwords with where they are used.

    Shipped 19 Oct 2026Open
  • Change risk score per publishLayer 7

    Every policy change scored with its reasons: services reached, sensitive assertions touched, effect on the grades, regression coverage and changed behaviour; in the publish dialog, on AI suggestions and against the previous version.

    Shipped 19 Oct 2026Open
  • Full chat and MCP coverage for Layer 7Layer 7

    Every Layer 7 page on the AI channels: gateway overview, security and health of one policy in context, diff, simulation, the health check report and what the scheduled sync found changed on the gateway. Read-only; publishing stays with a person.

    Shipped 19 Oct 2026Open
  • Estimates learn across migrationsNexus

    Every migration with finished flows and booked hours is kept in Nexus Estimates as one finished piece of work, so a project does not outweigh the others, and a new migration's estimate starts from what the earlier migrations really took instead of a generic assumption, unless one is set by hand.

    Shipped 26 Oct 2026Open
  • What standard products really takeNexus

    The CRM's project and milestone report loads as it is exported, and Estimates shows per product and product family what finished work took against the hours it was sold at, without treating sold hours as estimates.

    Shipped 30 Oct 2026Open
  • Quote documents read into the historyNexus

    A quote or estimate in Word, PDF, Markdown or text is read on our own server by a local model into a draft reference: the customer, the kind of work, fixed price or not, the estimated hours, the quoted price and the lines. People's names, mail addresses and phone numbers are left out, the hours really booked are added by a person, and a draft is not learned from until it is confirmed. On the Nexus Estimates page, over the API, and in chat and MCP as estimates.

    Shipped 25 Oct 2026Open
  • Quotes in the house styleNexus

    A saved estimate becomes two Word documents in the house style: the customer quote in Dutch or English (the scope, a fixed price rounded up with its milestones or the rate and expected budget, the assumptions, what is not included, the validity and a place to sign, never the buffer, the percentiles or a margin) and the internal pricing note for whoever approves the price (the range, the buffer, the margin per outcome, the history per kind of work and comparable finished work). On the Nexus Estimates page and over the API.

    Shipped 25 Oct 2026Open
  • Company knowledge baseNexus

    What we may say about ourselves, our solutions, our references and the texts that won them, in one base: every entry with its kind, class, sources and status, found by meaning (embedded on our own server) and by word. Unchecked statements are marked TO CONFIRM and never presented as fact, only a person approves, and confidential entries are for administrators. On the Nexus Knowledge page, over the API, and in chat and MCP as knowledge tools; the tender scorecard draws its picture of us from it.

    Shipped 31 Oct 2026Open
  • Nexus EstimatesNexus

    One calibrated estimator for every kind of work: managed integrations at a fixed price, migrations, gateway migrations, Layer 7 work, health checks, tender bids and consultancy. Finished work is kept as references (estimated or quoted versus the hours really booked, entered by hand or imported from a CSV with Dutch or English columns), and per kind of work and per customer the platform learns how far reality lands from the estimate and what fixed prices really earned per hour. New work is priced from lines with that history into a range and a price at P80, with comparable finished work beside it. In Nexus, over the API, and in chat and MCP as estimates.

    Shipped 25 Oct 2026Open
  • Robeeflow from the neutral modelNexus

    Every flow of an integration estate as a flow on our own Robeeflow hub, generated from the platform-neutral model, with every map as a mapping record carrying its XSLT and the trading partners in one configuration bundle the hub imports with a dry run first; every flow arrives inactive. Schedules become time triggers, a file read at the start a file trigger, called flows start by spawn; decisions, parallel paths with a join, try paths with error edges, maps, file writes, HTTP calls and mails become the hub's own nodes, and what the hub has no node for is named for a person to build. Every flow is proven against the hub's engine rules: on the real Adeptia export 164 of 165 flows are proven and all 289 mappings compile. On the migration Processes page as a bundle to download, over the API, and in chat and MCP as migration_robeeflow.

    Shipped 26 Oct 2026Open
  • Azure Logic Apps from the neutral modelNexus

    Every flow of an integration estate as an Azure Logic Apps (Standard) workflow, generated from the platform-neutral model: the trigger from its schedule, decisions as If actions, try and error paths as scopes, maps as XSLT, endpoints as built-in connections with their settings as app settings and no secrets in the files, calls as child workflows, scripts named for porting to .NET, and OpenAPI descriptions for API Management. Every workflow is proven, and loops are named rather than guessed. On the real Adeptia export 138 of 165 workflows are proven and all 289 maps compile. On the migration Processes page as a project to download, over the API, and in chat and MCP as migration_logicapps.

    Shipped 26 Oct 2026Open
  • Exit-readiness assessment for BoomiNexus

    How portable an integration estate is, read from the platform-neutral model so it serves any source the model is filled from: a portability score for the estate and every flow, the lock-in hotspots (scripts, the Java classes and platform functions called from maps, custom XSL, lookups, runtime values, connections), and Boomi, Azure Integration Services, MuleSoft, Apache Camel and Robeeflow compared on the map rules and scripts each carries by itself and the extra hours over the calibrated estimate at stated rates. On the real export: 85 out of 100, 143 of 165 flows portable. On the migration Overview page with a Word report, over the API, and in chat and MCP as migration_assessment.

    Shipped 25 Oct 2026Open
  • Trading partner portabilityNexus

    The EDI trading partners carried to a gateway other than Boomi: a Robeeflow configuration bundle with one partner per interchange id (stable record ids, so a second import updates; dormant partners suspended; no secrets, the hub issues each new partner its API key) and a gateway-neutral partner register with one row per partner, message and direction for any B2B gateway's bulk import, both with the onboarding plan and a note per partner. On the migration Partners page, over the API, and in chat and MCP as migration_partner_portability.

    Shipped 25 Oct 2026Open
  • Cross-platform regression harnessNexus

    The same messages replayed through any two runners and the outputs compared field by field, XML or JSON alike: the Adeptia stylesheet, the XSLT translated from the neutral model, a baseline recorded from Adeptia while it still runs, or a flow deployed on Boomi, Robeeflow, Logic Apps or Mule behind an HTTP endpoint (through the SSRF guard, its key used for the run only). Real messages are kept as cases per mapping, fields can be ignored, and every run is kept. On the migration Regression page, over the API, and in chat and MCP as migration_regression, where values from real messages are left out.

    Shipped 25 Oct 2026Open
  • Platform-neutral integration modelNexus

    Every Adeptia export read into one model that belongs to neither Adeptia nor Boomi: flows with their steps (receive, map, decide, try, call, script, send), maps with their rules in a neutral form (copy, constant, variable, concat, function, conditional, expression), schemas, connections with their secrets left out, endpoints, schedules, trading partners and scripts, checked for broken references. On the real export 165 flows, 289 maps and almost 15,000 rules with none broken. Downloadable as JSON from the migration Equivalence page, over the API, and in chat and MCP as migration_translate.

    Shipped 25 Oct 2026Open
  • Map translation engineNexus

    Each mapping translated from the neutral model into XSLT, JSONata or DataWeave: loops follow the target schema, fields inside a loop read relative to it, nested conditions and variables come through, and a rule that cannot be expressed is left as a TODO with the original, never guessed. The XSLT is proven by running it beside Adeptia's own stylesheet on samples, field by field; on the real export 88% of the rules translate to XSLT and about 2,500 compared fields are equal. On the migration Mapping and Equivalence pages, over the API, and in chat and MCP as migration_translate.

    Shipped 25 Oct 2026Open
  • Boomi licence optimiserNexus

    The Boomi licence the migrated estate needs, as generated and optimised: production and test connections and trading partners, with what each lever saves: retiring what is not migrated, one Disk connection per share instead of per directory, one trading partner per company, dormant partners left out. The contract's prices turn the counts into euros a year; none are assumed. On the migration Connections page, over the API, and in chat and MCP as migration_licence.

    Shipped 25 Oct 2026Open
  • Cut-over and parallel-run plannerNexus

    The switch of every wave planned: per flow which Adeptia trigger to pause before which proven-equal Boomi schedule starts, a parallel run on shadow folders for flows that write files or a direct switch with checks for flows that write to databases, mail or services, sub-processes before their callers, the rollback, and the wave's freeze and steps from its target date; risks that cross a wave's border are named. A runbook in Word in the house style. On the migration Waves page, over the API, and in chat and MCP as migration_cutover.

    Shipped 25 Oct 2026Open
  • Estimates learn from actual hoursNexussovereign

    Reference classes per product family and size show what comparable jobs really took (P50 and P80) beside every estimate line; each closed project gets a cause code chosen by a person, which a local model may draft from masked notes.

    Shipped 14 Nov 2026Open
  • Calibrated estimate in hours and moneyNexus

    The remaining migration and its total cost as a range in hours, euros and weeks, overall, per wave and per kind of work. The work is priced from what the platform has already generated and proven (process shapes by status, maps by their translation proof, scripts converted, drafted by AI or manual) plus the work around the flows (connections, EDI partners, testing, cut-over, hypercare, project management) at stated standard hours that can be changed per item; the team's ratio of actual over standard hours is learned from every completed flow with hours booked. P10, P50, P80 and P90, never one number, with the yearly Boomi licence and the three-year total beside it and the work breakdown as CSV. On the migration cockpit, over the API, and in chat and MCP as migration_estimate.

    Shipped 25 Oct 2026Open
  • EDI trading partner migrationNexussovereign

    The EDI estate as Boomi B2B with a partner-by-partner onboarding plan. Adeptia partners that share one interchange id (one record per legal entity of the customer) merge into one Boomi trading partner and the customer's own ids become My Company components; every partner gets its UNB control information, its communication channels and a processing rule per active relationship, and GLN ids are checked for their check digit. Active partners are dealt into four waves from a pilot to those that need information, with a note per partner on what changes for it and the plan in Word in the house style. On the migration EDI partners page, over the API with a zip of every component, and in chat and MCP as migration_b2b.

    Shipped 25 Oct 2026Open
  • Environment extensions and connection packNexus

    Every Boomi connection, operation and environment extension the migrated processes need, generated from the export: one shared connection per database, directory, FTP host, mail server, HTTP host, SOAP service and JMS provider, one operation per Adeptia source or target with its own settings, environment extensions for DEV, TST, ACC and PRD with production's values filled in, and every secret listed to set in Boomi (never taken from the export). The generated processes use the same names. On the migration Connections page, over the API with a zip of every component, and in chat and MCP as migration_connection_pack.

    Shipped 25 Oct 2026Open
  • Script converter for Adeptia services and custom classesNexussovereign

    Every Adeptia script a flow uses as a Boomi Groovy Data Process script. Inline BeanShell and Java are converted by rule (context variables, the document streams, service variables, logging, exceptions, Java forms Groovy reads differently), with every line that still needs a person listed and why. The compiled custom classes the export ships without source are read from their bytecode: a condition method is decompiled exactly, any other method gets its outline (variables read and written, calls, SQL, Adeptia platform calls). The platform's own model drafts the rest from the original, the conversion and the bytecode, and a local Groovy harness runs any script against a sample document. On the migration Scripts page, over the API with a zip of everything, and in chat and MCP as migration_scripts and migration_draft_scripts.

    Shipped 25 Oct 2026Open
  • Process and shape generator from Adeptia flowsNexussovereign

    Every Adeptia flow generated as a Boomi Process component on the developer's own canvas: sources and targets as connector shapes, mappings as Map shapes pointing at the generated maps, context variables as Dynamic Process Properties, exclusive gateways as Decision shapes in the gateway's own order ending in its default path, error events and exception scripts as Try/Catch, child flows as Process Calls. Each shape is mapped, partial (one setting to check) or manual (a pass-through saying what to build), and every process is proven against its flow: every step accounted for and every path leading to the same next step. On the migration Processes page and each flow page, over the API with a zip of every component, and in chat and MCP as migration_processes.

    Shipped 25 Oct 2026Open
  • Trigger and schedule converterNexus

    Every Adeptia trigger as Boomi runs it: Quartz crons converted to Boomi process schedules and proven to fire at the same moments over two weeks in their time zone, the Boomi pattern per event type (scheduled Disk, SFTP or Database polling, or a listener), with a CSV and AtomSphere ProcessSchedules payloads. On the migration Triggers page, over the API, and in chat and MCP as migration_schedules.

    Shipped 22 Oct 2026Open
  • Migrations into Layer 7 from MuleSoft and AxwayLayer 7

    MuleSoft API Manager APIs with their policies, Mule 4 applications and Axway APIM-CLI configurations converted into Layer 7 service policies in real gateway shapes, each source policy mapped, partial with its remaining step, or manual, and scored by the same analysis as any policy; secrets never copied. On the Layer 7 Convert page, over the API, and in chat and MCP as layer7_convert.

    Shipped 22 Oct 2026Open
  • Dutch health check reportLayer 7

    The Layer 7 health check report entirely in Dutch: findings, advice and control names from a fixed Dutch text set, and Dutch policy descriptions written on the platform's own model next to the English ones. Names from the Policy Manager stay as they are; monthly comparisons keep matching.

    Shipped 22 Oct 2026Open
  • Custom rolesPlatform

    Roles that limit an account to chosen tools (Layer 7, Migration, Tenders, Bastion, MEDDPICC, AI chat): enforced on every API route, in the chat tools and for Bastion as a non-admin member, with the menu following. Created on the Custom Roles page, given on the Users page, on the audit log; read-only in chat and MCP as access_custom_roles.

    Shipped 21 Oct 2026Open
  • Live Broadcom documentation refreshLayer 7

    The assertion catalogue the designer explains from is re-read from Broadcom techdocs on a schedule and per documentation version, compared with the catalogue in use (added, removed, renamed, moved, text changed, and buildable assertions that would lose their documentation link) and goes live only when an administrator approves it. In Layer 7 Settings, over the API, and in chat and MCP as layer7_vendor_docs.

    Shipped 21 Oct 2026Open
  • Mapping equivalence testerNexus

    Every Adeptia mapping and the Boomi map generated from its rules run on the same documents (two samples from the source schemas, plus any real message) and are compared field by field: equal, differs with both values and the input, not yet in the Boomi map, or not produced. A verdict per mapping, an overview with CSV, and the test on each mapping page; in chat and MCP as migration_equivalence.

    Shipped 20 Oct 2026Open
  • XSLT enginePlatform

    Our own XSLT 1.0 processor with XPath 1.0, in Go on our own server: the whole of both specifications, forward-compatible processing, the XPath 2.0 and EXSLT functions stylesheets for Saxon and Xalan use, Xalan redirect and result documents, extension functions in Go, and include, import and document() only through what the caller allows. Bounded in depth, size, steps and time; every error names its line and expression. All 289 stylesheets of a real Adeptia export compile.

    Shipped 20 Oct 2026
  • Run Adeptia mappingsNexus

    Run a mapping's own Adeptia XSLT on a document, or on samples built from its source schemas: Adeptia's date, Julian, variable, sequence and string functions run in Go, database and decryption calls are listed as not run. On the mapping page, over the API, and in chat and MCP.

    Shipped 20 Oct 2026Open
  • Boomi profile builder from Adeptia schemasNexus

    Every schema of an Adeptia export becomes the Boomi profile of its kind: XML, delimited and positional flat files, database statements, JSON from its sample, and EDI on the standard it includes. Keys are stable per schema and the generated maps refer to them, so maps import without re-pointing fields; what a schema does not say is listed for review. One zip with every profile, and chat and MCP see the coverage.

    Shipped 19 Oct 2026Open
  • Migration cockpit and weekly status reportNexus

    Every flow moves through analysed, generated, built, tested, accepted and cut over, one or many at a time, with blockers and their reasons and a target date per wave. Progress counts earned effort, so the weekly burndown and the forecast per wave move with the work. The weekly status report for the steering group downloads as Word with the burndown chart or goes out by mail every week; chat and MCP read the cockpit without names.

    Shipped 19 Oct 2026Open
  • Decommission workflow for dead flowsNexus

    Dead-flow candidates with their evidence, a named owner and a decision of migrate, retire or park; every decision recomputes the effort in scope, the waves and the connections no flow still needs.

    Shipped 19 Oct 2026Open
  • Migration tools in chat and MCPNexus

    The migration tool on every channel: imports, readiness, flows with their Boomi shapes, mappings and the wave plan as chat and MCP tools, so a consultant can ask which flows in wave 2 still have scripts to rewrite and an agent can work through a wave. Read-only; people named in an export stay out of the answers.

    Shipped 19 Oct 2026Open
  • Boomi Platform API push with dry run and rollbackNexus

    The components generated from an Adeptia export are created straight in the customer's Boomi account: a wave or chosen flows with the flows they call, their maps with both profiles and the connections and operations of their endpoints, in dependency order in a folder per wave, every bracketed id replaced by the id Boomi gives, every kind written the way Boomi's Build designer stores it (checked against Boomi's official component reference and a real account). A dry run checks first, every push keeps a ledger and a rollback removes exactly what it created; a corrected version can overwrite what is already in Boomi in place, keeping the ids, and a rollback restores the earlier version. EDI routing pushes the B2B side as Boomi's own design: a listener per standard, routes per partner and message in the processing group, inbound and outbound processes per relationship pattern, with their maps and EDIFACT profiles, and a cross reference table that sends each outbound message to the partner Adeptia would have chosen. The API token is sealed with the platform key. For administrators on the Processes page, in the AI chat and over MCP.

    Shipped 12 Nov 2026Open
  • Boomi read-back and migration overviewNexus

    What the pushes left in the customer's Boomi account, read back read-only and set against the plan and the push ledgers: per wave and kind the components in Boomi against those planned, per component whether it is as pushed, changed, renamed, deleted or moved in Boomi, not from the tool, never pushed or refused, with its Adeptia source, its Boomi link, deployments per environment, schedules and recent runs. A component renamed in Boomi is found by its id, so a later push never duplicates it. Read back by an administrator or daily; in the AI chat, over MCP and for read-only agents as migration_boomi_status, naming no one.

    Shipped 15 Nov 2026Open
  • MCP toolsPlatform

    More than 150 MCP tools across memory, documents, key-value, Spectre, the vault and Robeeflow, so agents and external developers share one surface.

    Shipped 6 Feb 2026Open
  • OpenAPI descriptionPlatform

    Every route is documented and served as one OpenAPI description with real response schemas, and a test fails when a route is missing from it.

    Shipped 26 Aug 2026Open
  • Configuration bundlesPlatform

    Export and import the platform's configuration as a bundle, with a dry run first.

    Shipped 15 May 2026Open
  • WebhooksNexus

    Outbound webhooks on platform events.

    Shipped 2 Apr 2026Open

Tenders workbench

A workbench for public tenders, from pipeline to questionnaire.

  • Tender WorkbenchTenders

    A pipeline, a bid/no-bid qualification, an answer library, questionnaires with blind-spot analysis, and DOCX export.

    Shipped 26 Sept 2026Open
  • Tender text libraryTenders

    Every text written for an award criterion, with its versions, the tenders it went into, the outcome and the assessor's score and motivation, searched by meaning (embedded on our own server) and by word, and shown as a graph. It learns from won and lost tenders, from scores and from definitive questionnaire answers. A quality check and a rewrite on our own local model work each text towards the best version, with invented claims flagged and a person deciding. In chat and MCP as tenders_library tools, and for the Librarian and Assistant agents.

    Shipped 31 Oct 2026Open

Sentinel workspace

The operator's workspace for deals, projects, ideas, reminders and the routing health of the AI fleet.

  • CRM and dealsSentinel

    Contacts, deals, MEDDPICC qualification, a forecast and a heatmap.

    Shipped 7 Apr 2026Open
  • Projects, ideas and remindersSentinel

    Projects with tools, an ideas backlog and reminders, all owner-scoped.

    Shipped 4 Apr 2026Open
  • Routing healthSentinel

    Which lane runs on which model, whether each chain has a local answer, and cascade health at a glance.

    Shipped 18 Aug 2026Open

Academy and Buddy

Learn the framework from inside the product, and keep a companion that grows with your use of it.

  • AcademyAcademy

    The laws, skills, flows, templates, glossary and manifest of the framework, readable in the product, with a leaderboard and XP progression. The agent gallery shows every agent that runs the platform with its role, permission profile, tools and recent work.

    Shipped 25 Mar 2026Open
  • Skill treeAcademy

    A skill tree that tracks mastery as agents and users complete work.

    Shipped 10 May 2026Open
  • BuddyAcademy

    One unique AI creature per user with rolled stats, evolution, missions, accessories and thoughts, generated on the local image model.

    Shipped 5 Apr 2026Open
  • All agents todayAcademy

    Every persona agent's day in one list on the Buddy's Today tab: its state now, today's runs, results and failures, and its next errand hour, your own marked. Readable in chat, over MCP and with a read-only agents:read key, which the agents' day pane in Claude Code uses.

    Shipped 2 Nov 2026Open
  • Living personasAcademy

    Each Buddy is the face of the platform agent its persona names: what that agent is doing now and did today, and errands at set hours that read and report for you (radar, deadlines, tender library, knowledge base, watched pages, the agent's work, a day report), filed as thoughts. Facts only, from records you may see; nothing is changed. The agent's real work earns the Buddy XP. In chat and MCP as buddy tools.

    Shipped 1 Nov 2026Open
  • Buddy house style painted locallyPlatformsovereign

    Every buddy creature is painted by the local FLUX model on your own GPU in one house style, a flat hand-drawn cartoon, and an admin action repaints the whole fleet after a style change.

    Shipped 27 Sept 2026Open

Design system and themes

One component suite, sixteen themes and a compliance score that can only go up.

  • Sovereignty UI component suiteDesign system

    Buttons, fields, cards, badges, modals, tables and more in one package with derived on-colour text and status families, used by every page.

    Shipped 19 Aug 2026Open
  • Pluggable themesDesign system

    Sixteen themes, one flat file each, switchable per user with a font picker and a theme audit.

    Shipped 19 Aug 2026
  • Token-only paletteDesign system

    Every colour is a design token; the palette lint, the undefined-token check and the ratchet run in the deploy so raw colours cannot come back.

    Shipped 24 Sept 2026
  • Design-system compliance gateDesign system

    A raw-primitive lint rule, an extended palette lint and a component index page, with a baseline that only ever rises.

    Shipped 27 Sept 2026Open

One transaction, one timeline

A message, the flow run it started, every node, every AI call with the model that answered, deliveries, acknowledgements, replays and cascade steps share one identifier and read as a single ordered story.

  • Transaction id contractObservability

    The platform's records about a message, a chat turn or a cascade carry the same transaction identifier, and the identifiers that already existed keep their meaning beside it. A copyable chip shows it on the timeline and the interactions page and lands on the same story. The hub's own rows and the remaining pages follow.

    Shipped 28 Sept 2026Open
  • Hub run recorder joined to the platform trace storeObservabilitysovereign

    The platform's trace store accepts a run from the integration hub as a tree of steps beside chat turns and cascades, and the timeline reads the hub's runs over the loopback. The hub sending its runs into that store is the step that follows.

    Shipped 28 Sept 2026
  • Replay lineageObservabilitysovereign

    A rerun on the hub links to the run it repeats, and a replayed message counts its replays, so an operator sees that a transaction was tried again. Dry runs and sandbox runs as recorded children follow.

    Shipped 28 Sept 2026
  • Unified transaction timelineObservabilitysovereign

    Paste any identifier and get one ordered list of everything that happened because of it on both sides, from the message received through nodes, AI calls, deliveries, acknowledgements, alerts, replays, cascade steps and audit rows, each with its source, state, duration, links and hashes. The same answer reaches the chat and the agent tools.

    Shipped 28 Sept 2026Open
  • Every AI tool call inside its traceObservabilitysovereign

    Every tool an AI calls, from chat, a cascade, the MCP bridge or an agent, is part of the trace of the conversation or run that asked for it, with the channel, the access verdict (allowed, refused by the role, refused by the agent's manifest), the agent and the size of the result. A model call the tool makes sits inside it. External agents' MCP calls are recorded too, which they were not before.

    Shipped 23 Oct 2026Open
  • History for AI batch runsObservability

    A long AI run keeps a record of itself. Every run has who started it, when, how it ended (done, or stopped by a restart) and its counts, and every item has its outcome, duration, error and the trace of its model call. First used by the Layer 7 description runs, on the Capabilities page and in chat and MCP.

    Shipped 23 Oct 2026Open
  • Audit log that tells the truthObservability

    The audit log records the status a request really got, so a request the role guard refused shows as refused instead of as a success, with how long it took and who acted. The platform's own calls, such as cascades and schedules, are recorded under the system identity instead of being left out.

    Shipped 23 Oct 2026Open
  • Prompt and response custodyObservabilitysovereign

    The text a model was given and gave back is kept only under the custody policy of its module. Mail and the vaults keep a hash only; chat keeps its text 90 days and everything else 30, with personal data and secrets masked before anything is written. The hash of the original always stays, so the ledger still proves what was asked once the text is gone. The interaction records are closed to everyone but superusers at the database.

    Shipped 24 Oct 2026Open
  • Decision ledger for AI permissionsObservabilitysovereign

    Every decision about what an AI may do is a row in a hash-chained ledger, allowed and denied alike. This covers a person's chat tool call judged by role and plan, an agent's call judged by its permission manifest, a person approving or rejecting AI work, and the kill switch. Each row is tied to the conversation it governed, and a day's chain can be verified on the AI Governance page, in chat and over MCP.

    Shipped 24 Oct 2026Open
  • Exact tokens and budgets per workspaceObservability

    Every AI call is counted with the provider's own token counts where it gives them (Ollama, NVIDIA, OpenRouter, Kimi, Gemini), its GPU or model time, and the customer workspace it worked for. Embeddings, image recognition and image generation are now counted too. Budgets limit tokens per day or month for a person, a task, a customer workspace or the whole platform. They warn the operators once at a chosen share and, when set to hold, refuse further calls until the period turns. Usage is shown per workspace, task, person or model on the AI Governance page, in chat and over MCP.

    Shipped 24 Oct 2026Open
  • Agent FloorObservabilitysovereign

    A live, read-only picture of every agent and cascade at work, laid out as a hub and spokes around the platform core with a live count per hub, the hand-offs between them with their grants, Robeeflow flow runs, the Standards Studio's progress and everything waiting for a person, built from real events only. It has its own page with a focus mode, follows light and dark mode and labels agents by name.

    Shipped 14 Nov 2026Open
  • Intellisense across the platformPlatformsovereign

    Texts recognise the agents, models, standards, laws, pages, code, versions and statuses they mention and show each as a coloured word, one colour per kind on every page (tickers, logs, run boards, chat, docs), a link in that colour where it leads somewhere, from a vocabulary assembled from the platform's own sources and terms people teach. The Documenter drafts new terms nightly for a person to confirm. Reusable as SmartText in the Sovereignty UI package; public showcase at /intellisense; on chat and MCP too.

    Shipped 14 Nov 2026Open
  • Quality signals on AI outputObservabilitysovereign

    What people do with AI output is tied back to the model call that made it, so every model and task gets a quality score from real use and not only from the bench. That covers a thumbs up or down under a chat answer, AI work approved or rejected in work review, and a Layer 7 description written again, with how much it changed. Scored per model, task or source on the AI Governance page, in chat and over MCP; any signed-in person can rate an answer and say why, and a weekly digest shows the ratings, the reasons and the change since the week before.

    Shipped 24 Oct 2026Open
  • What the AI readObservabilitysovereign

    Every piece of the platform's data a model is given is recorded by reference and hash, never copied. That covers a tool result in chat or for an agent, what an external AI fetched over MCP, a vault record, a memory recalled, a document, a Layer 7 policy being described. Each read carries the kind of data, what the call named, its size, and the conversation and person it was for. Administrators search it by person, kind or reference on the AI Governance page, in chat and over MCP; it answers which of my data the AI saw, and when.

    Shipped 24 Oct 2026Open
  • My AI activityObservabilitysovereign

    Every account has a page of what the AI did for it and with its data over the last 7, 30 or 90 days. That covers conversations, the tools used and how many were refused, the data the AI read by kind and reference, the AI runs it started, decisions about its requests, and tokens and cost. Open to every account whatever its role, and only ever its own. Chat answers the same with my_ai_activity, MCP with mcp_my_ai_activity.

    Shipped 24 Oct 2026Open
  • Model runtime telemetryObservability

    The local models are watched as a fleet every minute. That covers which model is loaded on the chat and the embedding instance and how much of it sits in video memory, the GPUs with memory, load and temperature, the queue with waiting calls and wait times, and tokens per second per model from exact counts. It also keeps the load, unload and spill events. A model that slips partly onto the processor, or a queue that keeps growing, alerts the operators. On the AI Governance page, in chat and over MCP.

    Shipped 24 Oct 2026Open
  • Replay an AI callObservabilitysovereign

    Any recorded model call can be run again with the same system prompt and input, on the same model or another, and the original and new answers are shown side by side. Use it to find out why an answer was wrong, and whether a newer local model would do better on real work. It stays within the custody policy, so a hash-only call (mail, vaults) cannot be replayed and a masked one replays on its masked text. The replay is itself a recorded model call. On the AI Governance page, in chat and over MCP, for administrators.

    Shipped 24 Oct 2026Open
  • Lineage or nothing, as the law saysObservability

    Modules that act outward or on personal data (mail, the vaults, Spectre, and every module its AI Act dossier rates high-risk) discard an answer whose lineage row cannot be written. Other modules spool the row, and a row that has not reached the ledger within 24 hours alerts the operators. The strict set is configuration the AI Governance page shows.

    Shipped 24 Oct 2026Open

Operations and deployment

The platform runs on one server you own, deploys in minutes and tells you what it needs.

  • Studio for the YouTube channelPlatformsovereign

    A private control room, open to one account, for an AI-assisted explainer channel. Overnight news becomes three pitches, and any lower-ranked topic can be pitched on request; a chosen pitch becomes a script on the local models, every claim is checked beside its source, and after two human approvals the platform renders the narrated video, two Shorts, thumbnails and a ready-to-upload package.

    Shipped 13 Nov 2026Open
  • Deploy that restarts only what changedOperations

    The deploy rebuilds and restarts the services that changed, fails when a stale binary is left behind, and preserves sovereign file ownership.

    Shipped 16 Sept 2026
  • DiagnosticsOperations

    A diagnostics page that checks the pieces the platform depends on and says which are missing.

    Shipped 14 May 2026Open
  • Status pageOperations

    A public status page for the platform.

    Shipped 15 Feb 2026Open
  • Plans and billingOperations

    PayPal subscriptions and comped plans, with feature access decided by role and plan.

    Shipped 16 Aug 2026Open
  • Dual-domain TLS transitionOperations

    The Go server terminates TLS itself for more than one domain and redirects to the canonical host.

    Shipped 19 Aug 2026
  • Compressed and cached responsesOperations

    Every text response is compressed and static assets are cached, with WebSocket upgrades left alone.

    Shipped 25 Aug 2026
  • Self-healing resumes after a restartOperations

    A cascade run killed by a deploy restart is recorded as interrupted rather than failed, the Self-Healing cascade re-fires at the next boot, and the deploy waits for running cascades before it restarts anything.

    Shipped 27 Sept 2026Open
  • Image service on the sovereign GPUOperationssovereign

    The local image service shares the GPU with the language models and decodes on the CPU, so it no longer runs out of memory beside them, and the buddy says when the service is not running instead of showing a placeholder.

    Shipped 27 Sept 2026
  • Fleet placement checkOperationssovereign

    A language model that does not fit beside the others on the graphics card used to be split onto the processor in silence, which costs minutes on every call. The fleet now checks where each model actually landed and reports a split model as a fault instead of letting it run slowly.

    Shipped 28 Sept 2026
  • Image service on demandOperationssovereign

    The local image service no longer holds sixteen gigabytes of video memory around the clock. It starts when an image is asked for and gives the memory back to the language models when it is not.

    Shipped 28 Sept 2026

Start with the framework and the governed workflow model, then add services where they help. See Pricing for hosted access, Documentation for how each piece works, and the Roadmap for what comes next.