Complete AI Training
Sign inGet my AI kit

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

AI agent for product designers

Design Decision Log and Rationale Agent

Keep a decision log where every design choice has an owner, options and linked evidence.

Document handlingAutomationCreativesProduct development
See how it worksGet access
Design Decision Log and Rationale Agent: what goes in, what the agent does and what you get

What it does

Six months after launch, a stakeholder asks why the button moved to the bottom, and nobody can find the reason. The same debate starts again. This agent collects decisions from design file comments, meeting notes and chat threads. For each decision it records what was chosen, which options were rejected, who decided and when. It links each entry to its evidence, such as a test result, analytics chart or research note. If a decision has no evidence, it asks the owner for it and checks again after a few days. It then checks that every entry has an owner, a date and at least one linked source. Anything still missing is shown as an open gap, not hidden. The design lead approves the log before it is shared. Edge case: two notes record opposite decisions, so the agent lists both and asks which one stands.

How it works

Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.

Start and resultWhat it doesA check on its own workWaits for your OKGoes back and retries
Yes, continueYes, continueApprovedNoNo 1 STARTS WHEN Sprint or design review ends 2 USES A TOOL Read comments, meeting notes and chat threads 3 DOES Extract each decision, its options and its owner 4 USES A TOOL Search research and test folders for matchingevidence 5 DOES Write or update the log entry and link the sources 6 CHECKS THE RESULT Does every entry have evidence and an owner? If not: ask the owner for evidence, wait, then searchagain. Back to step 4. 7 DOES Flag contradictions with earlier entries 8 CHECKS THE RESULT Are all contradictions resolved by an owner reply? If not: ask the owners which decision stands andre-extract the entry. Back to step 3. 9 YOU APPROVE Design lead approves the log 10 RESULT Published decision log
Read the steps as a list
  1. Sprint or design review ends
  2. Read comments, meeting notes and chat threads
  3. Extract each decision, its options and its owner
  4. Search research and test folders for matching evidence
  5. Write or update the log entry and link the sources
  6. Does every entry have evidence and an owner?If not: ask the owner for evidence, wait, then search again. Back to step 4.
  7. Flag contradictions with earlier entries
  8. Are all contradictions resolved by an owner reply?If not: ask the owners which decision stands and re-extract the entry. Back to step 3.
  9. Design lead approves the logThe agent waits here for your OK.
  10. Published decision log

How it decides

A statement counts as a decision when it names a choice and an owner. It counts as supported only when a linked source backs it.

  • Treat a statement as a decision only if it names a choice and a person who made it
  • Ask for evidence after 3 working days with no source
  • Mark an entry as superseded, never delete it, when a new decision replaces it
  • Escalate entries still unsupported after two requests to the lead

Make it yours

Every agent is a starting point. You choose these settings for your own situation.

  • Where decisions are collected from (comments, notes, chat)
  • Days before asking for missing evidence (default 3)
  • Log format and fields
  • Who counts as the approving lead
  • Which projects are covered

What keeps you in control

It always asks you first

  • Design lead approves the log before it is shared with stakeholders

Hard limits

  • Never invent evidence or a rationale that nobody stated
  • Never change who made a decision without confirmation

It stops when

  • Done: all entries have an owner, date and evidence, or are marked as open gaps
  • Stop: the source notes are missing for the whole period

Set it up

We guide you through the set-up, step by step

Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.

10 minto set it up in your AI
5 AIsChatGPT, Claude, Copilot, Gemini, Grok
  • One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
  • The agent then walks you through connecting your own data, one source at a time
  • A downloadable copy with the flow chart, the rules and the full guide
Get access to this agent

An example run

What happensAfter a checkout sprint the agent finds 11 decisions in notes and comments. Seven link to a usability test or funnel chart. Four have none. It asks the owners, and two reply with test links. On the recheck, two are still open and are shown as gaps. It also finds two notes about button placement that disagree, and the product designer confirms the later one. The lead approves the log.

More agents for product designers

Compliance · Automation

Prototype Usability Session Agent

Consistent, consent-respecting usability session records.

9 steps2 checks3 jobs
Quality control · Creating

Product Specification Example Agent

Specifications whose rules are tested by concrete examples.

7 steps1 check3 jobs
Analysis · Product development

Product Onboarding Friction Agent

Onboarding friction traced to its real cause.

8 steps1 check4 jobs
Creating · Quality control

Prototype Interaction State Agent

Prototypes that cover errors and recovery, not just the happy path.

8 steps2 checks3 jobs