Condens
A user research repository where AI stays in a supporting role. Condens transcribes sessions with timestamps, suggests tags by learning from your choices, clusters highlights by theme on a whiteboard and summarises a session with links back to transcript moments. The vendor owns one limit: the tool does not try to read emotion, tone or nonverbal cues. An MCP server opens published research to Claude or ChatGPT, read-only and within each user's permissions. Stakeholders read findings on a dedicated home page, or ask from Slack and Teams.
Strengths
- The vendor commits to not training models on your research data
- Hosting in Europe or the US on every plan, AI processed in the same region
- The MCP server cannot create, edit or delete anything, and only shows what the user is allowed to see
- Session summaries link back to the exact transcript moment
Limitations
- No free plan; the team plan jumps to €500 a month, paid yearly, for five contributors
- On Lite, auto-clustering and sentiment analysis only handle small datasets
- The MCP only reads published research, not your raw sessions
Best for
- A solo researcher or small team that wants a repository hosted in Europe
- Teams who want to query published research from Claude without granting write access