Sovereign AI Product · Documents

<GRIMOIR/>

Document
intelligence for your team.

GRIMOIR indexes policies, contracts, and operational files so your team asks in plain language and gets answers with sources, in minutes, not across three inboxes. Sovereign hosting. No public AI tools. EU-only data path.

Start the conversation Read the EU brief

// Sound familiar?

The answer exists. Nobody can find it.

Most organisations are not short of documents. They are short of a way to use them. GRIMOIR fixes that.

Someone left last quarter. Half the institutional knowledge left with them.

Your team spends twenty minutes finding a policy they know exists somewhere.

A client asks a question. Three people get called before anyone can answer it.

Sensitive documents are being summarised in public AI tools because it is faster.

// What it does

What GRIMOIR does for you.

GRIMOIR is built for businesses where answers need evidence, control, and accountability. Trusting the answer matters as much as getting one quickly.

01
Ask in plain language, get cited answers
Ask questions in natural language and get answers from your own files in seconds. Every answer cites the source document so your team knows exactly where it came from.
02
Expertise that outlasts the people who hold it
When someone leaves, their knowledge stays in the platform. GRIMOIR captures and indexes institutional expertise so turnover does not drain your regulatory or operational memory.
03
Across every document, every format
Policies, contracts, manuals, reports. Mixed formats across multiple systems. GRIMOIR indexes everything and makes it queryable from one place.
04
Cited, verified, and human-reviewed
Every answer comes back with sources from the original documents and is checked for accuracy before it is surfaced. Your legal and compliance teams can trust the trail.

// Where it operates

Who uses GRIMOIR.

GRIMOIR fits where the cost of a wrong answer is high, and the cost of exposing documents is higher.

01 / Domain

Legal

Contracts, precedents, and compliance documents. One missed clause costs more than the technology ever would.

02 / Domain

Finance

Regulatory filings, risk assessments, audit trails. The right answer needs to be findable in minutes, not a phone call away.

03 / Domain

Compliance

Policies change. Staff need the current rule, not last year's version. GRIMOIR keeps compliance knowledge current and queryable.

04 / Domain

Healthcare

Protocols, guidelines, and patient data rules. In healthcare, a wrong answer retrieved from the wrong document is not a recoverable mistake.

05 / Domain

Operations

Manuals, SOPs, and supplier contracts. The answer is in the system. Getting to it should not require three people and a morning.

06 / Domain

Government

Policy documents, correspondence, and regulations. Public-sector accountability starts with being able to find and cite the right information.

1 / 3

Your documents, queryable

Ask questions in plain language and get answers straight from your own documents. No third-party AI service. No pasting content into a public chatbot. No data leaving your business.

// How rollout works

From first domain
to whole business.

A staged rollout that earns trust one area at a time. Oversight, citation quality, and access rules stay in place at every step.

01 · First domain

Start where the value is clearest

We pick one high-value area of your business (contracts, policies, R&D notes) and tune the platform until citations are trustworthy. Trust is earned before scope expands.

02 · Ongoing oversight

Keep the rules visible

Access rules, review checkpoints, and verification settings stay visible to your legal, risk, and platform teams. The platform's behaviour is always understandable. Never a black box.

03 · Expand with confidence

Roll out across the business

Once the first domain is dependable, we extend GRIMOIR across more repositories and more teams. Same oversight, same identity rules, same human checkpoints, kept in place.

Want the EU-focused version? Read the brief: On-premise RAG platform (EU).

// Why not ChatGPT?

Your documents never leave your business.

When your team uses a public AI tool to summarise a contract or find a policy, your data leaves. It gets ingested, potentially retained, and processed on infrastructure you have no visibility into. GRIMOIR runs within the EU on your infrastructure. EU-only data path. Documents stay in your business. The same answers, without sending sensitive content to public AI tools.

EU-only data path

Documents and queries stay on EU-hosted infrastructure you control. Answers are generated entirely within your sovereign environment.

GDPR and EU AI Act ready

Access controls, audit trails, and human review checkpoints are native to the platform. Not bolted on before go-live.

Built in Malta for the EU

Engineered by a team working under EU law, for operators subject to EU law. The regulatory context is not an afterthought.

// Technical Specifications

For the engineers in the room.

What's under the hood, in case it matters to your platform team.

Model hosting: Open-source LLMs running entirely on your infrastructure. No external API calls, no data egress, no vendor handshake to answer a question your team owns.
Retrieval: Agentic retrieval with multi-step verification loops and citation-bound responses, designed for legal, finance, and regulated R&D.
Document ingestion: Platform-native support for PDF, Office, Markdown, HTML, and plain text from file stores and standard repositories. Proprietary DMS connectors scoped per engagement.
Deployment: Self-hosted on-premises, private cloud, or sovereign-cloud. Auto-scaling resources match query volume and corpus size.
Identity & access: OIDC and SAML integration so retrieval respects existing repository-level and document-level access rules. The same query returns different answers for different people, by design.
Audit & governance: Query logs, citation lineage, model versions, and operator actions are surfaced for legal, risk, and platform teams. Human-in-the-loop checkpoints applied where the cost of a wrong answer is high.
Custom models: Roadmap includes distilled, domain-tuned LLMs trained privately on your corpus and deployed alongside it.

// Next Step

Ready to ask anything?

Tell us about your documents, the rules around who can see them, and what your team needs to ask. We come back with a concrete rollout plan in plain language, with the technical detail your platform team needs.