RESPONSIBLE AI
Responsible AI built for governance & deal workflows
We add AI where it has the highest impact and where safety is guaranteed. Every capability we ship is useful, secure by design, and built to be trusted in regulated environments.The principles that govern our AI
“For companies like us, which provide board portals and data rooms, where trust, governance and compliance are fundamental, responsible use of AI is a necessity.”
Kristoffer Munthe Aasbø
Chief Product Officer, Admincontrol
Data security
Security and risk are assessed from the scoping stage of every AI feature.
Low tolerance for risk
Only the content an action needs is processed, and it never trains shared models.
Proven, not experimental
We only ship AI that is tested and compliant. Your board and your deals are not our test environment.
Transparency
Every AI output is labelled as such. Nothing happens out of sight.
Your data never touches the public internet
Every AI feature runs on infrastructure we control, designed so your documents stay isolated from start to finish.
Your documents and data
No public IPAI processing
No public IPLinked only by a private endpoint across a private network. Neither database can be reached from the internet.
Your data stays in Europe
Admincontrol Board Portal data sits in a private data centre in Europe. Admincontrol Data Room data is hosted in the EEA, in a dedicated virtual private cloud. AI processing runs in Europe.
Two isolated databases
Your documents and data sit in one database. AI processing happens in a separate one. Neither has a public IP address, and neither can be reached from the internet.
Connected privately
The two are linked only by a private endpoint across a private network. Content is tokenised and encrypted before it moves, processed inside the isolated environment, and returned by the same private path.
Controls built into every AI feature
Our AI features are designed around the same controls a board would expect from any critical system.
Governance by design
EU AI Act principles are embedded into the product, our processes and our controls from day one, not retrofitted later.
Transparency and explainability
Every AI output is clearly labelled, understandable, and reviewable by the people using it.
Human in the loop
AI drafts and suggests. People decide. The secretary, the counsel, the deal owner stays in control of the outcome.
Isolated architecture
Your documents and the AI processing run in separate, isolated databases, with no exposure to the public internet. See our AI architecture below.
Audit-ready by default
End-to-end traceability across the AI lifecycle, so you can show exactly what happened and when.
RESPONSIBLE AI
Consent-based, and always your choice
AI features are opt-in and apply only to organisations that consent to them and to our AI guidelines. They run in a virtual private cloud in European regions, with labelled outputs, human review and full auditability. Nothing is enabled without your agreement.
“When it comes to governance, trust is not the counterpart to innovation, but rather the foundation on which it rests.”
Kristoffer Munthe Aasbø
Chief Product Officer, Admincontrol
What our AI does today
Admincontrol Board Portal
AI Search
Semantic search across board documents, meeting packs and past decisions. Filter by folder or date, preview summaries in the search panel and find scanned files through OCR.
Learn more →AI Executive Summaries
One click on a document returns a structured summary, section by section. It is stored once, so everyone with access reads the same version. Available in eight languages.
Learn more →AI Agenda & Minutes Drafting
Drafts, edits, restructures and translates agendas and minutes. The secretary reviews and composes the final record.
Learn more →Admincontrol Data Room
AI Search
Natural-language search across every file in the data room, combining vector and keyword matching, with in-panel summaries and OCR for scans and images.
Learn more →AI Executive Summaries
One-click, section-by-section summaries of due diligence documents, stored once so every user with access reads the same version.
Learn more →FAQ
No. Your content is never used to train shared models. It is processed only to fulfil the request you make, inside infrastructure we control, and returned to you by the same private path.
Inside a virtual private cloud in European regions, with no public IP address and no route in from the public internet. Content is tokenised and encrypted before it moves, and it never travels over the public internet at any point.
No. Every AI feature is opt-in. It applies only to organisations that have consented to it and to our AI guidelines, so nothing is enabled on your portal or your data room without your agreement.
You are, and the product is built to make that workable. AI drafts and suggests; the secretary, the counsel or the deal owner reviews and decides. Every output is labelled, and every AI-assisted action is traceable in the audit trail.
EU AI Act principles are built into the product, our processes and our controls from the start rather than added later, covering transparency, human oversight, traceability and data governance. Our AI management system is designed in line with ISO/IEC 42001.
AI search in Admincontrol Board Portal and in the data room, AI drafting for agendas and minutes, and AI executive summaries. Anything still in development is described as planned, and never presented as available.