The Founder's Guide to Briefing an AI Project

Branched Team - 2026-05-06

A clear brief is the single biggest predictor of whether an AI project ships. The technology is rarely the bottleneck. Vague scope is. Here is the structure we use with founders to get from "we should do something with AI" to a project worth starting.

1. The decision or workflow you want to change

Not "use AI in marketing". Something specific: "respond to inbound enquiries within five minutes, 24/7, with a qualified next step." Name the moment in the business this is meant to change.

2. The number it has to move

Pick one. Response time, conversion rate, hours per client, utilisation, churn. Write the current baseline and the target. If you cannot baseline it, that is your first task - not the build.

3. The constraints

• What can the system never do? (e.g. quote a price, make a clinical recommendation, contact a client without review.)
• What data can it touch, and what is off-limits?
• What tools does it have to live inside? (CRM, helpdesk, email.)

4. The owner and the reviewers

One named owner accountable for the outcome. A short list of reviewers who sign off on what "good" looks like. No committees.

5. The first 30 days in production

Describe what week one looks like when it is live. Who uses it, how often, what they do when it gets something wrong. If you cannot picture this, the project is not ready to start.

6. What success looks like in 90 days

One sentence. Not a feature list - an outcome. "We are responding to 90% of enquiries within five minutes and have lifted booked calls by 20%."

What a good brief is not

• A list of features.
• A choice of model or vendor.
• A 40-page document no one will read.

One page. Six headings. A founder can write it in an hour. If you want a second pair of eyes on yours, talk to us - we review briefs before scoping any work.