Skip to content
Founderz: Responsible Use of AI home

in collaboration with Microsoft

All team tools

Team tool · Template

Single source of truth starter kit

Give your team’s key facts one home, one owner and a date, so people and AI find the same right answer.

Download

Blank template (.md)Fictional example (.md)

Principles involved

Overview

A single source of truth can start small, with the facts your team is asked about most: prices, policies, product details, contacts. Give each one a home, an owner and a date, and point your assistants there.

The five rules

  1. Every important fact has one home.
  2. Every home has an owner and a date.
  3. Point AI at the home.
  4. Drafts stay drafts until a person publishes them.
  5. Fix it at the source.

Principle 2 explains each rule.

First-week plan

  1. 01Day 1: list the ten facts the team is asked about most.
  2. 02Day 2: for each one, agree the official home and the owner.
  3. 03Day 3: add a version and a last-checked date to each home.
  4. 04Day 4: archive or delete the old copies you find, or mark them clearly as out of date.
  5. 05Day 5: ask your assistant the ten questions and check which document it uses for each.

Access review

Assistants usually see what the user can see. Microsoft’s guidance for Copilot deployments recommends reviewing access to overshared sites, and archiving inactive sites and deleting obsolete files so that answers stay current. Ask your IT team what applies to your tools.

The template

Each field says what to record. The fictional example shows what a finished record can look like.

Fictional example: the pricing facts of the sales team at Lumen, an invented software company.

01Information

Name the kind of fact: prices, a policy, product details, contacts.

Fictional example

List prices and discount rules for 2026.

02Official home

Say where the official version lives, with a link.

Fictional example

The Pricing 2026 page in the sales wiki.

03Owner

Name the person who keeps it current.

Fictional example

The revenue operations lead.

04Who approves changes

Say who can change the official version.

Fictional example

The sales director approves price changes, and the owner publishes them.

05Version and date

Say how the version and last-checked date are shown.

Fictional example

A version number and a last-checked date at the top of the page.

06Who can read it

Say which people and which AI assistants can read it.

Fictional example

All of sales and support. The work assistant can read it, and the public website chatbot can’t.

07Review rhythm

Say how often the owner checks it.

Fictional example

Monthly, and on every price change.

08When sources disagree

Say which source wins, and who resolves conflicts.

Fictional example

The Pricing 2026 page wins over slides and emails. The owner resolves anything unclear within two working days.

09Reporting a mistake

Say how anyone can report an error.

Fictional example

Comment on the page or post in #source-fixes.

10What relies on it

List the workflows, templates and assistants that use this fact.

Fictional example

Quote templates, the proposal assistant and renewal emails.

Sources

What each source establishes, and its limits. The practices and recommendations on this page are ours, and the facts come from the sources. See every source we use.

  1. Secure and governed data foundation for Microsoft Copilot: foundational deployment guidance Microsoft · May 6, 2026; updated August 18, 2026 · Product documentation Because Copilot can use anything the user can open, recommends finding overshared, ownerless and inactive sites, running access reviews, restricting broad sharing links, and archiving inactive sites and deleting obsolete files so answers stay current. Limits: Vendor guidance for administrators. Features depend on the license.
  2. Data, privacy, and security for Microsoft Copilot Microsoft · Updated August 18, 2026 · Product documentation States that Copilot only surfaces organizational data the user has at least view permission for; that prompts, responses and data accessed through Microsoft Graph aren’t used to train foundation models; and that generated responses aren’t guaranteed to be 100% factual, so users should use their judgment before sending output to others. Limits: Applies to that product and its commercial terms only.