Skip to content
Terry Mathew

AI Agents

Product Discovery AI

A structured workflow for early product research

A multi-agent prototype that organizes competitor research, customer problems, market sizing, risk review, and strategy into a structured report.

2025 · Working PrototypeCode ↗Demo ↗
Concept illustration: five layered paper research cards join through orange ribbons into one research brief
Generated concept illustration. Separate research cards combine into a brief; this is not a product screen or a real research result.

Inside the project

Explanatory diagram · Project workflow

Turn a product question into a research brief.

The workflow gathers signals, challenges assumptions, and combines findings. This diagram explains the project design, not a research result.

  1. 01

    Collect signals

    Research competitors and customer problems using search and Reddit tools.

  2. 02

    Challenge assumptions

    Sizing and risk roles examine the opportunity and its unknowns.

  3. 03

    Synthesize a brief

    Combine the research into a structured recommendation and roadmap.

This diagram explains the workflow. Generated reports need independent source review.

Separate the research roles

Each role examines a different part of the product question before synthesis.

Keep results open to review

A structured report helps people inspect the research. Its claims and estimates still need review.

Inspect the source ↗

01

The problem

Early product decisions need evidence about competitors, customer problems, market size, and risk. Collecting and combining that research can take time. A single research prompt can also miss important questions.

02

My role

I shaped the product direction and research workflow. I designed the system around separate research roles and a review step for the combined report.

03

The approach

The workflow separates research into five roles: market landscape, customer pain, opportunity sizing, risk review, and strategy synthesis. A Gradio interface presents the stages and their outputs. A quality-audit step flags claims that need stronger evidence. The report gives a structured view of the research for human review.

04

Challenges

Generic prompts produced shallow analysis. I added role-specific instructions and constraints. I also changed the pipeline to collect each role's output, so the final report could use more than the last result.

05

What I learned

Separate roles can broaden the questions a workflow examines. A review step can make weak evidence easier to spot. A structured report still needs a person to check its sources and conclusions.

06

Outcomes

The working prototype organizes research into reports about competitors, customer problems, opportunity size, and risks. The project has no verified measure of time saved or report accuracy. Reports need independent source review.

Tech stack

  • CrewAI
  • Python 3.12
  • OpenAI GPT-4o
  • Serper API
  • Reddit API
  • Gradio