Paste-and-go
Paste customer feedback directly on the home page and hit Analyze. Sign in with Google, and Mimir delivers ranked recommendations in about 60 seconds.
Projects
Create a project to organize your sources and analysis. Each project has its own insights, recommendations, and chat history.
Uploading sources
Drag-and-drop files onto the inbox or chat panel — or click to browse. Supported formats: .txt, .csv, .pdf, .png, .jpg, .webp.
What happens next
Mimir extracts key entities from your sources, clusters them into themes, and generates ranked recommendations — typically in 15–20 seconds for a single source.
Every project has an overview page that serves as your home base. When you return to a project, Mimir generates a briefing that summarizes where things stand and suggests what to do next.
Briefing
An AI-generated executive summary that catches you up on your project — what the data says, what's changed, and what action to take next. Generated automatically when you return to a project.
Signals
Key metrics at a glance: strongest theme (with trend direction), top recommendation, data freshness, and decision momentum (how many recommendations are being explored, built, or shipped).
What's changed
After re-analysis, a summary of how your themes evolved — new themes that emerged, themes that strengthened or weakened, and themes that resolved. Helps you track how your understanding shifts as you add more data.
Impact / effort matrix
An interactive scatter plot showing all your recommendations positioned by impact and effort. Click any dot to jump to that recommendation. Quick wins (high impact, low effort) stand out immediately.
Every project has an AI chat panel grounded in your sources, insights, and recommendations. Ask questions about your data, get strategic advice, or brainstorm what to build next.
File attachments
Click the paperclip icon or drag-and-drop files directly into the chat. Mimir will process them as new sources and incorporate them into future analysis.
URL awareness
Paste a URL into the chat and Mimir will read the page content automatically. Works with bare domains too — no need to type https://.
Type / in the chat input to see all available commands.
Insights
AI-synthesized themes ranked by severity and frequency. Click any theme to see evidence quotes traced back to the original sources.
Re-analysis diffs
When you re-run analysis after adding sources, Mimir shows what changed — new themes, strengthened themes, weakened themes, and resolved themes — so you can track how your understanding evolves.
Recommendations
A ranked list of what to build next. Sort by quick wins, highest impact, lowest effort, status, or AI-recommended order. Filter by status, impact level, and effort level.
Status workflow
Track each recommendation through New → Exploring → Building → Shipped (or Rejected). Status persists across re-analyses.
Impact projections
AI-generated charts showing projected metric improvement over 6 months — baseline versus projected, so you can see the expected trajectory.
Analytical markers
Colored highlights in recommendation rationale that surface cross-source contradictions, root cause fixes, and primary metric movers. These appear automatically when Mimir has enough sources to cross-reference (3 or more).
Live refinement
Use /refine to debate a recommendation with the AI. Challenge the priority, narrow the scope, or ask for the MVP version. The rationale text updates live as you talk — no page refresh needed.
Sharing
Create shareable links with optional expiration (7 or 30 days). See who has viewed the link and revoke access anytime. Shared pages include the spec and agent tasks.
Every recommendation can generate two outputs: a build spec and a set of agent tasks. Together they close the loop from “what to build” to “building it.”
Build specs
A structured markdown document covering the problem, proposed solution, user impact, and implementation approach. Written for humans — product managers, designers, or engineering leads who need to review and approve the work.
Agent tasks
Discrete, copy-pasteable task blocks formatted for coding agents. Each task describes exactly what to build, where in the codebase, and what “done” looks like. Paste them directly into Claude Code, Cursor, Copilot, or any AI coding tool.
Copy & download
Copy specs or agent tasks to clipboard with one click, or download them as markdown files. Share them with your team or feed them into your development workflow.
Push recommendations directly to GitHub as issues — no copy-pasting or context switching.
Connect your GitHub account
Link your GitHub account from any recommendation detail page. This is account linking, not login — your existing Google sign-in stays the same.
Connect a repository
In project settings, select which GitHub repo receives issues. Each project can connect to a different repo.
Create issues from recommendations
On any recommendation's Spec or Agent Tasks tab, click “Create issue” to push the spec content, agent tasks, and a mimir label to your connected repo. The button transforms into a link to the created issue.
Capture Slack threads as Mimir sources without leaving Slack.
Connect your workspace
One-time setup. Go to Settings and click “Connect Slack.” This installs the Mimir bot in your Slack workspace.
Capture threads
In any Slack thread, type /mimir or mention @Mimir. You'll see a summary of the thread and a dropdown to pick which project to import it into — or create a new project right from Slack.
What gets captured
The full thread is saved as a source with participant names, timestamps, and reactions preserved. A link back to the original Slack thread appears on the source in Mimir.
Batch and analyze
Mimir doesn't auto-analyze when you capture a thread. Capture as many threads as you want, then run analysis from the web UI when you're ready.
Connect Intercom, Linear, and Notion to import conversations, issues, and pages directly as sources. No copy-pasting, no exports — just one-click imports from the tools you already use.
Connect a provider
Go to Settings and click “Connect” on any provider card. You'll authorize Mimir to read data from your account via OAuth. Tokens are encrypted at rest and never shared with third parties.
Import data
On the Sources page, click a provider's icon to browse available items (Intercom conversations, Linear issues, or Notion pages). Select what to import, and Mimir saves them as sources. Deep links back to the original items are preserved.
What gets imported
Intercom: Full conversation transcripts with customer messages and team replies. Linear: Issue title, description, and comments. Notion: Page title and full markdown content.
How it works
OAuth is handled by Nango, a secure token proxy. When you connect a provider, Nango stores the access token (encrypted) and proxies API requests on your behalf. Mimir never sees your raw token — we send fetch requests to Nango, which adds the token server-side.
Disconnect anytime
Revoke access from Settings. This deletes the stored token and stops all API access. Previously imported sources remain in your project.
Mimir automatically extracts knowledge from your sources and chat messages, organizing it into 7 categories: company profile, competitive landscape, user segments, goals & constraints, product state, terminology, and metrics.
Confidence scoring
Each knowledge entry has a confidence score based on how many independent sources support it — so you can see what Mimir is most and least sure about.
Manual management
View and edit all knowledge entries in Settings. Delete anything that's wrong or outdated.