Step 01
Gather Public Data on a Key Client
Collect the minimum useful public source set and record title, date, URL, and topic for each item. Exclude private messages, meeting notes, confidential deliverables, and personal data.
IntermediateProduct
How to Create a 'Synthetic Client' with a Custom GPT for Faster Project Iteration
Create a clearly labeled synthetic client from authorized public material to surface likely questions and priorities before a meeting, while keeping real client feedback authoritative.
Try this with your AI assistant
Download the skill for this workflow. Version 1.0.0.
Source episode · 43:00How I AI: Building a Personal AI Wellness Coach to 'Feel 25 in a 40-Year-Old Body'
with Claire Vo
Lucas describes a synthetic version of a time-constrained client built from her public articles, podcast appearances, and presentations, which his team uses to anticipate questions before meeting with her.
Before you start
What you need
- A client’s public articles, talks, interviews, and presentations
- Permission and account rules for the chosen AI workspace
- A dated list of known priorities and project constraints
- A real client meeting or feedback checkpoint
What you’ll make
A source-linked internal critique that identifies likely questions, alignment, contradictions, and unknowns without claiming to speak for the client.
Tools used
- ChatGPTVisitChatGPT
OpenAI conversational AI
The workflow
Follow the sequence once, then adapt the prompts, checks, and handoffs to your own setup.
5 steps
Step 02
Create and Train a Custom GPT
Create a Custom GPT that analyzes the material as a synthetic model of likely priorities, not as the person. Require citations, uncertainty, and third-person language.
Example prompt
You are a clearly labeled synthetic client-analysis assistant based only on the attached public sources about [person]. Do not claim to be them, invent quotations, or issue approvals. When reviewing work, identify likely priorities and questions, cite the source behind each, distinguish evidence from inference, surface contradictions, and mark unsupported topics as unknown.
Step 03
Query the 'Synthetic Client' for Internal Feedback
Submit the current proposal or design and ask for likely questions, points of alignment, tensions, missing evidence, and a counterargument. Keep the output attached to its sources.
Example prompt
Analyze this draft against the public source set. Return likely questions, alignment, concerns, missing evidence, and one strong counterargument. Cite each attributed priority and state confidence. Do not predict approval.
Step 04
Iterate Internally Based on AI Feedback
Revise only where the critique reveals a real issue, and keep alternative choices visible. Record which assumptions need the real client’s answer.
Step 05
Present the Refined Work to the Real Client
Present the work to the real client, ask the recorded questions, and update the project brief from their actual feedback. Treat discrepancies as useful calibration, not as errors to hide.
What good looks like
- Every attributed priority traces to public source material.
- The assistant labels inference and refuses unsupported opinions.
- Outputs stay internal and are never presented as client approval or quotation.
- The real client confirms, rejects, or changes the assumptions in the next feedback cycle.
Build your next product with ChatPRD
Turn an idea into a PRD, user stories, and a plan.
Runbook notes
How to recover when the loop fails and where human judgment helps.
If it goes sideways
- Confidential client material enters a model without authorization
- Remove it, follow the client and company response process, and rebuild the source set from explicitly approved material.
- The output reads like a real quote or decision from the client
- Use third-person language, cite sources, label it synthetic, and prohibit generated approvals or signatures.
- Older public statements conflict with the current brief
- Prefer dated current sources, surface the conflict, and ask the real client rather than selecting one silently.
- The team optimizes for the simulation and stops exploring alternatives
- Ask for counterarguments and unknowns, retain divergent options, and use the real meeting to test the assumptions.
Related workflows
intermediateDataMeta Muse
How to Use AI for Hobby Improvement and Data Analysis for Golf & Chess
Connect data from your hobbies (like golf scores or chess games) to an AI agent to get automated analysis, visual dashboards, and actionable coaching insights. Turn raw data into a story that helps you improve.
Open
beginnerPersonalMeta Muse
How to Create a Personalized Shopping Catalog from Your Instagram Saves Using AI
Turn your unstructured Instagram 'Saved' collection into a fun, personalized shopping catalog. This workflow uses an AI agent to organize your saves and present them in a nostalgic, engaging format like the SkyMall magazine.
Open
advancedContentMeta Muse
How to Revive an Old Blog or Digital Archive as a Personal AI Podcast
Resurrect your old digital content, like a forgotten blog from Blogger or Tumblr, by using an AI agent to find, consolidate, and transform it into a daily personal podcast. Reconnect with your past self in a new, listenable format.
Open
Start shipping better products.
Join 250,000+ product managers who use ChatPRD to write better docs, align teams faster, and build products users love.
Free to start · 14-day Pro trial · SOC 2 Type II
