Sources are the raw material Mimir works with. The more you add, the more patterns Mimir can find.
Text
Paste directly into the chat or upload .txt files. Customer interviews, meeting notes, feedback threads — anything text-based.
CSV
Upload .csv files for structured feedback — survey exports, NPS data, support ticket dumps.
Upload .pdf files. Mimir extracts text content and processes it like any other source.
Images and screenshots
Upload .png, .jpg, or .webp files. Mimir uses vision to read text from screenshots, app store reviews, Figma exports, and more.
Slack threads
Mention @Mimir in any Slack thread to capture it as a source — complete with participant names and timestamps. See the Slack integration guide for setup.
When you add a source, Mimir classifies it (interview, feedback, analytics, notes) and extracts structured signals — pain points, feature requests, observations, and metrics. Each signal is traced back to the original source, so you can always verify where an insight came from.
After extraction, Mimir looks across all your sources to find patterns. Signals that appear in multiple sources become themes, ranked by severity and frequency.
Sources also feed Mimir's knowledge base — business details, user segments, competitor mentions, and terminology are extracted automatically and used to sharpen future analysis.