Create a project-specific AI workspace from approved internal context and traceable customer research, then use it to draft product artifacts without treating scraped conversations or model summaries as an unbiased source of truth.
Amir builds a project-specific AI brain from kickoff decks, PRDs, product pages, and Reddit research, asks Claude to analyze recurring themes with counts and quotes, spot-checks the evidence, and adds reviewed artifacts back to the project.
A project workspace with governed context, a reproducible customer-feedback dataset, source-linked themes, and reviewed product artifacts that can be updated as the evidence changes.
Anthropic AI assistant
OpenAI conversational AI
Step by step
Follow the sequence once, then adapt the prompts, checks, and handoffs to your own setup.
5 steps
Create one workspace for one initiative. Load the minimum approved kickoff, PRD, product, and research context, and label every file with its source, date, owner, and confidentiality level.
Adopt an 'everything is text' mindset. Print web pages or convert slides to PDF to broaden the scope of what your AI brain can learn from.
Define the research question before generating collection code. Ask Claude to compare supported sources, terms, access, expected coverage, rate limits, and fields, then build against the permitted API or export without embedding credentials.
I need evidence about [research question] from [candidate sources] for [date range]. Compare supported access methods, terms, expected coverage, rate limits, privacy risk, and cost. Recommend a permitted source and minimal fields. Then propose a reproducible collector that reads credentials from environment variables and records source URL, timestamp, query, and stable record ID.
Store raw records separately from analysis. Compute basic counts with code, then ask the model to group recurring topics, retain rare counterexamples, and cite the exact record behind every quote and claim.
Analyze this versioned dataset for [question]. Use the provided deterministic counts. Return theme, count, denominator, affected source segments, representative record IDs and quotes, contradictory record IDs, and confidence. Do not call this sample representative of the market.
Spot-check the top themes, quotes, counts, and counterexamples against the raw records. Correct unsupported findings and compare the sample with interviews, support, sales, or product data before using it in a decision.
You summarized that [insert a specific finding from the AI's analysis]. Please provide 3-5 direct quotes from the uploaded data that support this conclusion.
This verification step is crucial for ensuring the AI isn't hallucinating and that your product decisions are based on real data.
Configure the workspace to act as a candid product partner while separating facts, source-backed findings, assumptions, and recommendations. Save reviewed artifacts back as dated, versioned context rather than relying on chat history.
Use only the approved workspace files. For every response, separate source facts, research findings, assumptions, open questions, and recommendations. Cite the file and date for factual claims. Challenge my proposal when the evidence conflicts, and say when the workspace lacks current evidence.
Turn an idea into a PRD, user stories, and a plan.
After the steps
How to recover when the loop fails and where human judgment helps.
Recover
Keep building
Design a weekly meta-workflow that analyzes your interactions with your AI assistant to make it smarter. This system learns your writing style from edits, suggests new automations for repetitive tasks, and helps you filter valuable AI techniques from hype.
Create a daily AI workflow that summarizes yesterday's meetings, identifies urgent tasks, and proactively learns internal company terms it doesn't understand. This keeps your AI assistant's knowledge current and highly relevant to your work.
Build a recurring AI task that automatically gathers updates from your calendar, Slack, and notes to suggest weekly priorities and prepare you for upcoming meetings. Start every Monday with a clear, comprehensive plan without the manual effort.
Join 100,000+ product managers who use ChatPRD to write better docs, align teams faster, and build products users love.