Home / Services / AI systems / Internal knowledge assistants

Service and operations

Answers from your own documents, with the source attached.

Help staff find answers across policies, tickets and meeting notes. The assistant cites the file and passage, respects the permissions of the person asking and flags questions the documents cannot answer.

Book a consult →All eleven systems

Who it is for

Teams that need this workflow.

Firms where the answer exists in a document nobody can find in under ten minutes

Operations teams onboarding people faster than the documentation can be read

Departments whose policy, procedure and precedent live in three systems and a shared drive

What it does

Internal knowledge assistants, by the job.

Index

Indexes with permissions

Documents, tickets, wikis and chat are indexed inside your tenancy, and the permissions of the person asking are carried through to what can inform their answer.

Answer

Answers with citations

Every answer quotes the passage it came from and links to the document. A reader can check it in one click instead of trusting it.

Refuse

Refuses when the corpus is silent

A question with no source in your documents gets "not in our documents" and a suggested owner, rather than a fluent guess.

Update

Stays current

When a document changes, its entries update. When a session or a process creates a new document, it is indexed before the day ends, with no wait for a nightly rebuild that may not run.

How it is built

The build, in order.

The pilot covers one defined workflow. We agree its success metric and kill threshold in writing before the build; the steps below show how this system is developed and evaluated.

Census the corpus

Which systems hold the knowledge, who may see what, how stale each source is, and which questions your team asks most. The questions your team asks set the scope.

Build the index and the gates

Retrieval with permissions, citation to the passage, and the refusal rule. An evaluation set of real questions with known answers is written with your team before anything is tuned.

Pilot with one team

A team that asks the most questions uses it for a month. Every wrong or missing answer is logged, and each one is either a corpus gap or an index defect, fixed as such.

Widen and maintain

More sources, more teams, and a weekly report of unanswered questions, which is the most useful list a documentation owner ever gets.

In production at Strataigize

We run this ourselves.

Our own company runs on this. A set of knowledge graphs covers our code, our knowledge base, our operating files and this website, one connector serves all of them, and every working session queries the graph before it opens a file. When a session changes a file, it updates the graph before it closes, which is what keeps the assistant current without a nightly rebuild. A graph that cannot answer something it should is logged as a defect and repaired.

How our agency uses AI agents

Governance

Controls and human review.

Answers respect the permissions of the person asking; a document they cannot open cannot inform their answer
Every answer cites its source, and a question with no source in the corpus is refused rather than guessed
Your documents are indexed inside your tenancy and never used to train a shared model

The rest of the posture is published on trust and security: client-granted access you can revoke, credentials in managed stores, an AI governance summary aligned to the NIST AI Risk Management Framework, our subprocessors, and a DPA on request. We do not hold SOC 2 and we say so there.

Price

Prices in the open.

Pilot builds $25,000 to $60,000 fixed, production retainer from $8,000 per month. Every AI build starts with a $7,500 AI Opportunity Audit over two weeks that is credited toward the build if you proceed, and a pilot is one bounded workflow on your real data over 30 to 90 days with a written success metric and a kill threshold.

The full price list is on the pricing page. For the factors that change a quote, read how much AI agents cost in 2026.

Case studies

The work behind it, documented.

Strataigize Marketing

How Our Marketing Agency Uses AI Agents to Automate Busywork

40+ Estimated client tasks queued monthly7 days A week of industry news, read in minutes daily

Read the case study →

Strataigize Marketing

Our AI Productivity Journey: Research & Case Study

+25% Quality uplift in the UK AISI workplace trial4+ hrs Saved weekly by heavy users, US Fed survey

Read the case study →

Browse every case study →

Questions

Internal knowledge assistants, answered.

What is an AI knowledge base?

A searchable index of your own documents that an assistant answers from, with the passage cited, instead of a general model answering from what it learned on the public internet. The value is in the permissions, the citations and the refusal rule.

Is this retrieval-augmented generation?

Retrieval-augmented generation is the usual mechanism, and it is the easy part. What decides whether the assistant is trusted is what we spend the time on: permissions carried through to the answer, citations to the passage, a clear refusal when the corpus is silent, and an index that updates when a document changes.

Which systems can it read from?

Google Drive, SharePoint and OneDrive, Notion, Confluence, Slack, HubSpot, ClickUp, help desks, and any system with an API or an export. Read access is granted by you and can be revoked by you.