Dovetail
A user research repository with an AI layer on top. You pour in interviews, calls, reviews and tickets; Dovetail transcribes, highlights against your own tags, clusters into themes and answers in chat with citations you expand to check the source. Chat runs on Claude, and you can scope it to a few chosen interviews instead of the whole project. A panel shows which interviews it read. On top come agents whose write actions wait for a human sign-off, Digital Twins that answer like a customer segment from what it said in research, and an MCP server. No free plan: the entry tier is billed yearly for five users.
Strengths
- Every chat answer carries citations you expand without leaving the conversation, and a panel lists the interviews it read
- You can query three chosen interviews out of forty instead of the whole project
- Customer data is not used to train the vendor's models
- Agents can wait for a human sign-off before writing to Salesforce or Slack
Limitations
- No free plan: the entry tier costs $4,800 a year for five users, plus credits
- AI cannot be turned off workspace by workspace
- Theme clustering works best in English; French is supported, no more
- Automatic redaction of personal data and choice of model provider are reserved for the top tier
Best for
- Research teams piling up interviews, calls and tickets who need to find an exact quote
- PMs who have to show where a conclusion comes from, excerpt in hand