> Documentation index: https://usertold.ai/llms.txt

# About UserTold.ai

UserTold.ai connects product delivery to what users say and do. Many teams can already turn a clear issue into code quickly. The harder upstream problem is deciding which issue is worth solving and preserving enough evidence for another person or agent to verify that decision.

## What we build

UserTold runs interviews inside a product. With a participant's consent, a study can combine conversation with product use, navigation, clicks, and page context. Completed interviews become source-linked Evidence: reviewable moments with quotes, observed behavior, context, confidence, and uncertainty kept distinct.

Related Evidence can support a draft Finding. A project-aware human or agent reviews its sources and synthesis; review does not create external work. Product triage decides whether a reviewed Finding becomes delivery work. After delivery, the Evidence stays connected to the result so later Evidence can prompt recurrence review instead of starting the research history over.

## What we believe

Recording comes before interpretation. Evidence should remain inspectable independently of a summary. Uncertainty should be visible, and generated interpretation should never be presented as if a participant said it. Product automation should make failures and recovery paths understandable rather than hide them behind magic.

UserTold is for builders, founders, product-minded engineers, and teams already using coding agents. It is not a replacement for consent, product judgment, or project-aware review. It is the evidence those decisions start from.

## Learn more

- [How UserTold works](/how-it-works)
- [Research methodology](/docs/methodology)
- [Security](/security)
- [Privacy](/privacy)
- [Support](/support)
