Knowledge base AI systems
Your documents, contracts and history turned into a source your team can ask questions of.
- Engagement
- Fixed-scope build, optional retainer if you want us maintaining the corpus
- Typical timeline
- 7 weeks
- Starts with
- A map of where your documents live and who is allowed to read each store.
Eleven years of work sits in Drive, SharePoint, a wiki nobody has updated since a reorganisation, and a shared inbox. The answer to most internal questions already exists, finding it requires knowing it exists and remembering who wrote it, and whatever you build must never show a new analyst the board pack.
Institutional knowledge that no longer walks out of the door.
What we build
Concrete artefacts, handed over and documented.
Connectors to the places documents genuinely live, syncing on a schedule rather than as a one-off import that goes stale in a month
Permissions mirrored from the source system and checked at the moment of asking, so the assistant can never surface a file the person could not open themselves
Citations naming the document and section with its last-modified date, because a confident answer from a 2019 policy is worse than no answer
Conflict handling when two documents disagree, which shows both rather than silently choosing
A report of the questions people asked that your documents could not support
What changes
Institutional knowledge outlasts the people who leave with it
Onboarding stops consuming your senior team’s afternoons
The gaps in your documentation become a ranked list instead of an assumption
How it runs
- Week 1
01Sources and permissions
Where documents live, who may see each store, and which archives are safe to include at all.
- Weeks 2–4
02Ingest and ground
Connectors, indexing and citation handling, with permission checks tested by trying to break them.
- Weeks 5–7
03Department pilot
One department uses it daily, and the unanswered-question report becomes their documentation backlog.
Chosen per project. Named here so you can see the shape of it.
- pgvector
- Postgres
- OpenAI
- Microsoft Graph
- Google Drive API
Questions we get asked
Tell us the outcome, not the tooling.
Send us the situation you are in. We will tell you which discipline it belongs to, what we would do first and what it costs, including when the answer is to wait.