<✦>aicodedesign.Start building

THE HUMAN + AI WORKFLOW

Know when to ask AI.
Know what to own.

A practical map from problem framing to a working prototype. At each stage, see what context to provide, where AI helps, what you decide, and what evidence lets you move forward.

Start with your project, not a blank chat.

Use the planner to write a focused handoff for your current stage. Then keep the actual decisions and results in your Project Passport.

Build my AI handoffCreate my test checklist →

01 · DESIGNER + AI

Frame the problem

Give AI: An anonymized brief, observed difficulty, constraints, and the decision you need to make.

AI can help: Turn your notes into candidate problem statements, identify gaps, and suggest questions.

You decide: Choose the real audience and verify the problem with evidence. AI-generated personas are hypotheses, not research participants.

Produce: One user, one task, a scope boundary, and observable success criteria.

Before moving on: Can you distinguish observed evidence from assumptions and name what you will not build?

02 · DESIGNER + AI

Explore alternatives

Give AI: The accepted brief, constraints, and examples of what users need to understand.

AI can help: Generate different flows, sketch directions, content structures, and tradeoffs.

You decide: Select a direction for a stated reason. Check whether familiar patterns fit this audience and context.

Produce: Two or three alternatives and your decision record.

Before moving on: Have you compared materially different approaches instead of cosmetic variations?

03 · DESIGNER + AI

Specify screens and behavior

Give AI: Design references, tokens, sample records, and interaction rules.

AI can help: Translate your design into component states, draft microcopy, and implementation requirements.

You decide: Own hierarchy, accessibility, scope, and rules. Explain which visual decisions must be preserved.

Produce: A state inventory: empty, loading when relevant, success, invalid input, no results, and failure.

Before moving on: Does every visible action have a defined result, error response, and data boundary?

04 · DESIGNER + AI

Build one complete interaction

Give AI: The approved behavior specification and the relevant source files, without credentials.

AI can help: Implement a small change, explain the relevant files, and propose verification steps.

You decide: Review the change, run it, inspect the result, and keep a recoverable source checkpoint.

Produce: One working flow from input through result, plus a record of the checks performed.

Before moving on: Can you reproduce success and failure yourself, and explain where the data lives?

05 · DESIGNER + AI

Debug with evidence

Give AI: Expected versus actual behavior, reproduction steps, relevant code, and sanitized error messages.

AI can help: Trace the reported failure, suggest competing causes, and propose a minimal repair.

You decide: Reproduce the issue, review the evidence, and rerun the original and adjacent cases.

Produce: A small fix and a regression check tied to the original failure.

Before moving on: Did the original failure stop occurring without breaking a nearby interaction?

06 · DESIGNER + AI

Test with people

Give AI: The task and prototype limitations; afterward, actual anonymized observations.

AI can help: Draft neutral task scripts and organize anonymized observations into themes and open questions.

You decide: Recruit appropriate consenting testers, observe real behavior, and decide what the evidence supports.

Produce: Observed problems, separate interpretations, a prioritized change, and a recheck.

Before moving on: Are findings traceable to actual observations? Did you record help given and limitations?

07 · DESIGNER + AI

Release and tell the story

Give AI: Actual checks, known limitations, the source checkpoint, and your personal contribution.

AI can help: Draft release notes, a maintenance checklist, and a case-study outline from supplied evidence.

You decide: Decide readiness, confirm ownership and data handling, verify the deployed URL, and approve public claims.

Produce: A release decision, maintenance owner, working link, and honest portfolio story.

Before moving on: Do public claims match implemented behavior and observed results? Is rollback documented?

TOOL 01 · AI HANDOFF BUILDER

Give your AI partner
a clear next step.

This rules-based tool formats your context into an editable prompt. It does not call an AI service, evaluate your design, or save your entries. Your text stays in this tab; export before leaving. Use anonymized information when sharing the output elsewhere.

TOOL 02 · ACCEPTANCE CHECKLIST BUILDER

What should you test?

Choose the closest prototype type for a starter test plan. Adapt expected results to your actual rules. These are checks to perform, not passed tests or a readiness score.

A worked example: the missing filter

Imagine a reader saves a book but cannot find it after changing its status. Start by reproducing the behavior. If the active filter intentionally excludes it, the problem may be discoverability or feedback rather than failed saving. Give AI the reproduction steps and current rules, then ask for competing explanations. You choose whether to improve the status message, expose the active filter more clearly, or change the interaction.

Implement one change, repeat the original task, and observe a tester. Record the actual result in your Passport. This is a hypothetical teaching example; it is not a report of research performed on the reference app.

Useful tools already here

The Project Passport keeps your brief, decisions, debugging record, test observations, and case study together. The working reading-list reference provides editable source and real local interactions. Use this workflow map to connect those tools to your next action.

When to slow down

AI can propose research questions, but it cannot turn invented answers into user evidence. It can draft a repair, but you still need to reproduce and recheck the failure. It can outline a launch plan, but customer data, permissions, transactions, and recovery need appropriate implementation and review. Match the scope of your claims to what you actually verified.

Download the workflow mapExplore the toolkit →