The quality of your brief determines the quality of what you get back. A vague brief produces vague estimates, scope creep, and disappointing results. A precise brief gets you accurate pricing, realistic timelines, and a product that matches your expectations. Here is exactly what to include.
Section 1: Business context (2-3 paragraphs)
Tell the agency who you are and why this project matters. This is not fluff — it directly affects architectural decisions.
- Company overview: Industry, size (employees, revenue range), target market. "We are a 50-person B2B SaaS company selling project management tools to mid-market construction firms."
- Current pain point: What specific problem are you solving? Be concrete: "Our support team handles 800 tickets/month, 65% are repetitive questions about billing and feature usage. Average response time is 6 hours."
- Business goal: What does success look like in numbers? "Reduce average response time to under 2 minutes for common questions, and free up 3 support agents to handle complex technical issues."
Section 2: Scope definition
This is where most briefs fail. Be specific about what you want built:
- Core functionality: List the 3-5 things the product must do. Not 20 features — the essential ones. "The AI agent must: (1) answer billing questions using our knowledge base, (2) check order status via our API, (3) escalate complex issues to human agents with conversation history."
- Channels: Where will this live? Web chat on your website? Email? WhatsApp? Slack? Each channel adds integration work.
- Integrations: List every system the product needs to connect to. CRM (which one?), payment processor, knowledge base, ticketing system, email. Be specific about versions and APIs available.
- User roles: Who uses the system and what can each role do? "Admins configure the agent. Support leads review escalated conversations. Customers interact via web chat."
Section 3: Technical constraints
Every project has constraints. Stating them upfront prevents wasted time:
- Existing tech stack: What systems are already in place? Cloud provider, database, authentication system, deployment pipeline. The agency needs to work within (or migrate from) your existing infrastructure.
- Data sensitivity: What compliance requirements apply? GDPR, HIPAA, SOC 2? Are there restrictions on where data can be stored or processed?
- Performance requirements: Response time expectations, uptime requirements, concurrent user load. "Must handle 500 simultaneous conversations with under 3-second response time."
- Security requirements: SSO integration, encryption requirements, audit logging needs, IP restrictions.
Section 4: Budget and timeline
Agencies need this information to propose a realistic solution. Hiding your budget does not help you — it results in proposals that do not match your resources:
- Budget range: Provide a range, not a single number. "$15,000-$30,000 for initial build" is more useful than no budget information.
- Timeline: When do you need this live? Is there a hard deadline (conference, regulatory requirement) or is the timeline flexible?
- Ongoing budget: What can you spend monthly on maintenance, hosting, and LLM API costs?
Section 5: Success metrics
Define how you will measure whether the project succeeded:
- Primary KPI: One number that defines success. "AI agent resolves 65%+ of support tickets without human intervention."
- Secondary KPIs: 2-3 supporting metrics. "Average response time under 30 seconds. Customer satisfaction score above 4.2/5. Support team handles 40% more complex cases per week."
- Measurement method: How will you track these? Existing analytics? New dashboards? The agency should build measurement into the product.
Section 6: Reference materials
Attach everything relevant:
- Current workflow documentation or process maps
- Sample data (anonymized if sensitive)
- Screenshots of current tools being replaced or augmented
- Links to competitors or similar products you admire
- Brand guidelines (if customer-facing)
- API documentation for systems you need to integrate
Common mistakes in briefs
- Describing the solution instead of the problem. "Build us a chatbot with GPT-4" is a solution. "Reduce support response time from 6 hours to 2 minutes" is a problem. Let the agency propose the solution — they may have a better approach than what you have imagined.
- Listing 50 features. An MVP needs 3-5 core features. Everything else is Phase 2. Feature overload leads to budget overruns and delayed launches.
- No success metrics. Without measurable goals, there is no way to evaluate whether the project was worth the investment.
- Ignoring ongoing costs. The build is 40-60% of total first-year cost. Hosting, LLM APIs, maintenance, and iteration make up the rest. Budget for the full picture.
A well-written brief saves weeks of back-and-forth and thousands of dollars in scope changes. At N40, we offer a free 30-minute briefing session where we help you structure your requirements before any commitment. The result is a clear scope, accurate estimate, and realistic timeline.