From User Interviews to Linear and GitHub Issues
A coding agent can implement “add CSV export” perfectly and still solve the wrong problem. UserTold turns reviewed Interview Evidence into Findings that preserve the affected user, workflow, observed problem, expected outcome, uncertainty, and source links before product triage or delivery.
UserTold groups related Evidence into draft Findings in the backlog. A human or project-aware agent checks the source moments, synthesis, current product context, and uncertainty. Review does not create external work; the receiving team decides during product triage whether a reviewed Finding becomes delivery work.
Start with a reviewed problem
Do not hand a literal request directly to a delivery agent:
“Add CSV export”
-> inspect the user, workflow, behavior, expectation, and alternatives
-> review source-linked Evidence
-> verify the current problem
-> review Finding
-> push to Linear or GitHub
Use Turn Feedback Into Work Worth Building first when the Evidence is still incomplete or a draft Finding has not been reviewed.
The handoff path
- Run interviews against one focused study.
- Extract Evidence from available transcript and observed-behavior sources.
- Review Evidence by interview, type, confidence, and source moment.
- Group related Evidence into a draft Finding.
- Check the grouping against current project and product context.
- Correct, split, defer, dismiss, or mark the Finding reviewed.
- Explicitly send a reviewed Finding to Linear product intake or GitHub delivery.
Draft and reviewed are research states of the same Finding, so the synthesis stays connected to the Evidence that supports it. Generated delivery context remains an interpretation; it is not presented as something the participant said.
What should a reviewed Finding contain?
| Question | Reviewable context |
|---|---|
| What user-experience pattern may be meaningful? | Finding synthesis and linked Evidence |
| Who experienced it and where? | Interview and available page context |
| What did the user say or do? | Quote or observed facts with source moments |
| What did the user expect? | Participant explanation or clearly labelled interpretation |
| Why do these moments belong together? | Grouping rationale, counter-evidence, and uncertainty |
| Does the problem still exist? | Current project and product review |
| Is it ready for delivery? | Ready status after human or project-aware agent review |
This gives an engineer or coding agent enough context to investigate without treating participant or model interpretations as root-cause facts.
Choose the delivery tracker
Use Linear when product work is triaged and prioritized there. Use GitHub Issues when accepted delivery work lives closer to the repository. In either tracker, keep the UserTold Finding link so reviewers can return to the source Evidence.
Linear completion can resolve currently linked Evidence in UserTold and make future similar evidence available for recurrence review. GitHub does not sync completion back to UserTold, and recurrence is not causal proof that a shipped fix succeeded or failed.
Continue after handoff
Choose the MCP integration, CLI reference, or dashboard for the handoff path your team already uses. After a linked Linear issue is completed, continue with Track Resolved Evidence.