← Back to all guides

From client chat to clear brief

How to extract structured requirements and draft clean briefs from raw client communications.

Who is this guide for?

This guide is for account managers, freelancers, and service providers who receive unstructured briefs, loose chat logs, or raw audio transcriptions from clients and want to convert them into a structured brief draft.

What this guide helps with

The workflow extracts project objectives, deliverables, constraints, and deadlines from unstructured text. Crucially, it highlights omissions and ambiguities that you need to resolve with the client.

Before you begin

Confirm that you understand the current workflow, required inputs, decision rules, and where human review must remain. If the process is not yet documented, start by mapping it.

Prompt / materials

Use the following template structure to process chat transcriptions and draft your project brief:

Act as a requirements analyst. Parse the provided client communication history and extract a structured project brief. Input Chat/Transcript: [Paste raw anonymized chat messages or meeting notes here] Deliver the output in this structure:
Continue reading the prompt
1. Core Objective: What is the main problem the client is trying to solve?
2. Scope of Work & Deliverables: Specific items or features requested.
3. Timeline & Constraints: Deadlines, technical limits, or design requirements mentioned.
4. Ambiguities & Open Questions: List at least 3 things that are unclear or missing from the input that require follow-up discussion.

Safe-Use Rules & Guidelines

  • Anonymize Client Details: Always remove company names, client names, contact details, email addresses, or specific server credentials before pasting chat logs into public AI tools.
  • No Decisions: The AI only organizes and structures raw text. It does not finalize contracts, make legal commitments, or make engineering scope decisions.
  • Professional Review: Final review of the brief remains your responsibility. Make sure to double-check that the AI did not misinterpret the client's intent or invent deliverables.

Example input

Here is an example of an informal client request that you can use as input:

Client: Hey, we need a simple contact page on our site. Needs a map and a form.
Consultant: Sure. Do you want it to save submissions somewhere?
Client: Just email them to us. Need it by next Friday if possible. Oh and we need custom styling, no standard templates.

Short example output

Sample format of the output prepared by the AI assistant for your review:

1. BRIEF SUMMARY
- Client wants to add a custom contact page with a map and a form by next Friday.

2. CONFIRMED FACTS
- Form submissions sent via email.
- Custom styling required (no standard templates).

3. TO VERIFY
- Which email address should receive submissions?
- What map service is preferred?

4. PRIORITY QUESTIONS FOR CLIENT
- What email address should receive submissions?
- Do you have a preferred map provider (e.g. Google Maps)?

5. SAFE DRAFT REPLY
"Hi, thank you for reaching out. We can build this custom contact page. To help us outline the details and timeline, could you clarify..."

What to check in the output

Before sending or utilizing the output from the AI assistant, verify the following:

  • Accuracy of facts: Check that the AI correctly captured the client's request details without inventing new features.
  • Tone: Ensure the drafted response aligns with your brand voice and sounds natural.
  • Scope: Confirm that the working brief only lists requested features.
  • Promises and commitments: Ensure that no pricing, strict deadlines, or final delivery commitments are accidentally made or implied.

Related guides

Need help implementing this guide in your workflow?

We design custom prompt sets, API integrations, and train teams on safe and efficient AI usage.

Help with implementation Explore Services