How to Brief an AI Project: Scope, Budget, Timeline

Knowing how to brief an AI project is the skill of creating a structured document that describes your problem, your current situation, and your desired outcome so that AI agencies can deliver comparable, well-founded proposals. Without a brief, you get three quotes that solve three different problems — none of them yours.

Yet most SMBs start their search for an AI partner without a brief. They call three agencies, tell a slightly different story each time, and receive proposals they cannot compare. The result: decisions made on gut feeling instead of evidence.

This article gives you the seven sections that belong in every strong AI brief, the most common mistakes, and a comparison of briefing approaches so you can start writing today.

Why a Brief Makes the Difference

Writing an AI project brief takes 3-4 hours. That investment pays for itself in three ways:

Comparable proposals. Every agency answers the same questions, with the same scope and the same frame of reference. You compare apples to apples instead of concepts to implementations.

Lower project costs. Agencies that receive a clear brief spend fewer hours on discovery and interpretation. Research by the Standish Group shows that projects with a well-defined scope come in 28% cheaper on average than projects that start from a vague idea.

Faster turnaround. No weeks of back-and-forth over scope. The agency reads your brief, asks targeted follow-up questions, and delivers a proposal within 5-7 business days.

The 7 Sections of a Strong AI Brief

1. Problem statement

Describe the problem you want to solve, not the solution you think you need. An agency that understands your problem can propose better solutions than you would come up with yourself.

Wrong: "We want an AI chatbot on our website." Right: "Our 4-person support team spends 60% of their time on recurring questions about order status, returns, and opening hours. That's 96 hours per month of work that requires no specialist knowledge."

2. Current process metrics

Provide concrete numbers about your current situation. Without a baseline, no agency can calculate a realistic saving, and you can't measure whether the project succeeded after the fact.

Include:

• Hours per week spent on the process
• Number of transactions, tickets, or invoices per month
• Error rate in manual work
• Cost of the current process (hours x hourly rate)

Don't have these numbers yet? Read our article on whether your business is ready for AI first — it helps you establish the right baseline.

3. Desired outcome and success criteria

Define what success looks like in measurable terms. "Work more efficiently" is not a success criterion. "Reduce incoming order processing time from 4 hours to 30 minutes" is.

Ask yourself:

• What percentage of the work should be automated?
• How many hours per week do you want to save?
• What is the maximum error rate you accept?
• When is the project successful — after 1 month, 3 months, 6 months?

4. Budget indication

Yes, you need to state a budget range. Many buyers refuse because they fear overpaying. But the opposite is true: without a budget indication, you receive proposals that don't match your reality, wasting your time and the agency's.

Not sure what's realistic? Our AI costs overview for SMBs gives you a concrete reference frame per project type.

Always state a range, never an exact number. That gives agencies room to propose solutions at different levels.

5. Timeline and deadlines

Specify when you want the system live, and whether there are hard deadlines (seasonal peaks, compliance requirements, contract renewals). An agency that knows your timeline can propose realistic phasing.

6. Technical context

Describe which systems the AI project needs to integrate with. Be specific:

• CRM system (Salesforce, HubSpot, Pipedrive, etc.)
• ERP or accounting software
• Communication platforms (email, WhatsApp, phone)
• Current website or webshop and its platform
• Available data and where it is stored

The more technical context you provide, the more accurate the proposal. Integrating with Salesforce requires different effort than integrating with a spreadsheet.

7. Decision-making process

Tell the agency who decides, who is involved in the evaluation, and what the timeline looks like. Nothing is more frustrating for both parties than a proposal that sits on a desk for two months because the director is on holiday.

Good vs. Bad: a Brief Compared

The 5 Most Common Mistakes

1. Prescribing the solution. "We want a chatbot with GPT-4 integrated with our Zendesk." Maybe a chatbot is the right solution, maybe not. Describe the problem and let the agency determine the approach. That's what you pay them for.

2. Not giving a budget indication. Agency A quotes EUR 8,000 for a basic solution, Agency B quotes EUR 45,000 for an enterprise system. Both are honest — they just didn't know what you could spend.

3. Not quantifying the current situation. "We spend a lot of time on X" is not useful. "We spend 20 hours per week on X, split across 3 team members" gives an agency the basis to calculate ROI.

4. Contacting too many agencies at once. Three is the optimal number. With five or more, the quality of your evaluation drops, and agencies notice when you send a mass inquiry — the better ones may decline. Read our guide to choosing an AI agency for a structured selection process.

5. Skipping internal alignment. You send out a brief, receive three proposals, and then your finance director says the budget isn't there. Align internally on budget, timeline, and decision authority before you start.

From Brief to Selection

Your brief is ready. Here is the process that follows:

Week 1: Send the brief to 3 selected agencies. Give a deadline of 10 business days for the proposal.

Week 2-3: Agencies ask follow-up questions. Schedule a call or visit of maximum 1 hour per agency. This is also your chance to assess communication style and working methods.

Week 3-4: Proposals received. Score them using a matrix based on your success criteria. Compare total costs including maintenance — not just the build price.

Week 4-5: Call references, optionally hold a second meeting with your top 2, make your decision.

The entire selection process takes 4-5 weeks. That sounds long, but it prevents months of frustration from choosing the wrong partner. For the full context on working with external AI specialists, read our complete guide to hiring AI consulting — the pillar article in this series.

Start Your Brief Today

A good AI project brief is not a bureaucratic document — it's the foundation of a successful project. The seven sections in this article give you a complete framework that any agency can work with immediately.

Open a blank document, work through the seven sections, and within an afternoon you will have a brief that gets you comparable, well-founded proposals. That is the difference between a considered investment and an expensive gamble.

Want to discuss your AI brief or get a no-obligation consultation straight away? We're happy to think along — no strings attached.