Gather the knowledge
We connect the document stores and systems the assistant should use and decide who may see what.
AI Implementation · AI Assistants & Agents
Answers from your own knowledge — with the source attached.
Internal AI assistants that answer from your own policies, manuals and records with citations, and AI agents that triage email and update systems — with access control and human oversight.
What is an AI knowledge assistant?
An AI knowledge assistant is a chat interface that answers staff or customer questions using your organisation’s own documents and data — policies, manuals, procedures, contracts, product information and records. It uses retrieval-augmented generation (RAG): the system first finds the relevant passages, then the AI writes an answer from them and cites its sources, respecting each user’s access rights.
What assistants and agents are used for
- Answering HR and policy questions
- Technical manuals and procedures for technicians
- Customer service and product questions
- Searching contracts, leases and tenders
- Triage and drafting of email replies
- Updating CRM and ticketing records
- Summarising meetings, reports and cases
- Explaining dashboards and alarms in plain language
How it works
From first survey to working system
Index and ground
Content is split, indexed and kept in sync so answers come from current documents, with citations.
Add actions carefully
Agents are given specific, logged tools — create a ticket, update a record — with approval steps where the risk warrants it.
Evaluate continuously
A test set of real questions is scored before launch and after every change; feedback from users improves it over time.
Why it pays back
What you get
- Answers in seconds instead of searching folders
- Consistent answers, with sources
- Knowledge that stays when people leave
- Routine email and admin handled automatically
Technology
Standards and platforms
ClaudeOpenAIAzure AIRAG & vector searchMicrosoft 365 / SharePointPythonEvaluation harnesses
Where it applies
Industries we deliver this for
Insights
Further reading
FAQ
Frequently asked questions
Will the assistant make things up?
Grounding answers in retrieved documents, requiring citations and instructing the model to say when it does not know sharply reduces this. We test for it explicitly before launch.
Can it respect who is allowed to see which documents?
Yes. Retrieval is filtered by the user’s permissions, so the assistant only uses documents that person could open themselves.
What is the difference between an assistant and an agent?
An assistant answers questions. An agent can also take actions in other systems — such as updating a record or sending a reply — within limits you set.
Also in AI Implementation
Next step
Tell us what’s slowing you down.
Describe the process, the system or the site. We’ll come back with a clear, practical view of what can be automated — and what it would take.