Step 01
Use Forks for Safe Exploration
Mark the last working prototype as the baseline. Before a meaningful feature or layout change, create a fork so the experiment cannot damage the version you already trust.
IntermediateEngineering
How to Systematically Iterate and Debug AI-Generated Prototypes
Keep AI prototypes easy to explore and recover by naming a stable baseline, creating a fork for each direction, and separating diagnosis from implementation when a change fails.
Try this with your AI assistant
Download the skill for this workflow. Version 1.0.0.
Source episode · 18:38How I AI: Colin Matthews' Workflows for Prototyping with Your Brand's UI
with Claire Vo
Colin and Claire show baseline prototypes, named variations, side by side forks, and checkpoints that make experiments easy to compare and reverse.
Before you start
What you need
- A working prototype worth preserving
- A prototyping tool with forks, copies, or version history
- One feature or design question to explore
What you’ll make
A stable baseline with clearly named variants and a repeatable path for diagnosing failed edits.
The workflow
Follow the sequence once, then adapt the prompts, checks, and handoffs to your own setup.
4 steps
Step 02
Adopt a Clear Naming Convention
Name every fork for the question it explores, such as “Baseline: profile page,” “Variation 1: inline edit,” or “Variation 2: split layout.” Keep one direction per fork so comparisons remain useful.
Step 03
Diagnose AI Errors Before Solving
When an instruction fails, stop issuing variations of the same fix. Ask the agent to identify the relevant component, state, event, and reason the current behavior persists, with no code changes yet.
Example prompt
I tried to remove [element or behavior], but it still appears. Inspect the implementation and explain the root cause. Name the relevant files, component state, and event flow. Do not edit any code yet.
Step 04
Implement the Solution
Review the diagnosis, then ask for the smallest targeted change. Test it in the current fork and create a checkpoint as soon as the behavior works.
Example prompt
Apply the smallest change that fixes the root cause you identified. Remove [element or behavior] and any state or event handler used only by it. Keep unrelated layout and behavior unchanged, then run the relevant checks.
What good looks like
- The baseline stays usable throughout the experiment.
- Each variant represents one clear direction.
- A teammate can compare versions without reconstructing the chat history.
- Failed edits can be reversed without rebuilding the prototype.
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
- The working version is lost during an experiment
- Return to the last named checkpoint or fork and create a fresh variant before trying the change again.
- Variants are impossible to distinguish
- Rename each one with the feature, hypothesis, or layout it tests and archive abandoned branches.
- The agent repeats a failed edit
- Ask it to explain the relevant files, state, and root cause without writing code, then issue a separate implementation request.
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
