Use case

AI assistant for company documents

The team asks in plain language and the AI answers based on the company's manuals, policies and procedures, always showing which document the answer came from.

The scenario

How it works today, without automation.

The answer exists, but it is in an 80-page PDF, in a folder few people can find. So the team asks the most senior person, who stops what they are doing to answer for the tenth time.

  1. 01New hires take weeks to learn procedures that are already written down.
  2. 02Everyone answers differently, and the official procedure isn't followed.
  3. 03The most experienced people spend the day answering questions.
  4. 04Outdated manuals sit next to new versions and nobody knows which one applies.
How it works with AI

The flow, step by step.

  1. 01

    Documents organized

    Manuals, policies and procedures gathered into one knowledge base.

  2. 02

    A person asks

    In the internal chat, Teams, Slack or WhatsApp.

  3. 03

    AI searches the documents

    Finds the passages that answer the question.

  4. 04

    Answers with the source

    The answer comes with the document name and the passage used.

  5. 05

    No answer, it says so

    If it isn't in the documents, the AI says it doesn't know and points to who to ask.

Sample proposal

What would be delivered.

Deliverables

  • Knowledge base with the company's documents, updated automatically.
  • AI assistant in the channel the team already uses.
  • Answers with the source shown and access control by department.
  • Report of the most asked questions and the ones left unanswered.
// connected systems
  • Google Drive
  • SharePoint or OneDrive
  • Notion or Confluence
  • Microsoft Teams
  • Slack
  • WhatsApp

Stages

  1. Discovery

    Which documents go in, who can see what and where the team will ask.

    2 to 3 days
  2. Knowledge base and assistant

    Organizing the documents, search and channel integration.

    2 to 3 weeks
  3. Testing with the team

    A group uses it day to day and rates the answers.

    1 week
  4. Go-live and tuning

    Rollout to everyone and monitoring of the questions.

    1 week
Typical timeline
3 to 5 weeks

The final timeline is set in the proposal, after discovery.

Frequently asked questions

Questions about this use case.

Can the AI make up an answer?

It is configured to answer only from the documents and to show the source. When the answer isn't there, it says it doesn't know.

Does each person only see what they are allowed to see?

Yes. Access follows department permissions: an HR document, for example, only shows up for people who have access to it.

When a manual changes, does the AI update?

Yes. The knowledge base is synced with the documents folder, and the new version applies from the next sync.

Are company documents used to train third-party AI?

No. Documents are consulted at the time of the question, with providers and contracts that don't use the data for training.

Want to see how it would work in your company?

You explain the problem. We build the solution. A straight answer, no generic proposal.