How It Works
UserTold.ai closes the loop between user behavior and product decisions.
1) Define the study
Create a study with goals, script segments, and screener rules. This defines who qualifies and what the interviewer should learn.
2) Run interviews
Participants enter through your screener and complete guided sessions. The AI conducts voice interviews live — observing behavior, asking follow-up questions, and stepping in when someone gets stuck. We capture voice, transcript, and interaction context.
3) Extract evidence
The platform turns raw sessions into structured signals:
- struggling moments
- desired outcomes
- workarounds
Each signal links back to source evidence.
4) Create tasks
Signals cluster into actionable tasks with titles, priority, and context. Tasks can be pushed to your tracker (for example GitHub) from the app.
5) Measure impact
After shipping changes, rerun interviews and compare signal rates. This verifies whether the fix improved user outcomes.
Agent and API friendly
You can operate the loop from the UI, CLI, or MCP tools with JSON-first outputs.
Learn more
- Quickstart — zero to first interview in 10 minutes
- Studies — configure the interview script
- MCP Integration — connect via MCP for agent workflows
- CLI Reference — command-line reference
- For Agents — agent-specific integration guide