Getting started

Create a workspace, add your first source, and see what Mimir finds — in about a minute.

What is Mimir

Mimir is an AI-native product management tool. Tools like Cursor and Claude Code help teams write code, but the harder problem is figuring out what to build. Mimir closes that gap.

The core loop: add customer interviews, product usage data, and feedback to your workspace → Mimir reads your sources, finds patterns, and surfaces what deserves attention → you get evidence-backed recommendations to guide your next move.

If you're a founder, product manager, or engineering lead who owns product decisions, Mimir is built for you.

Paste-and-go activation

The fastest way to see Mimir in action. No setup, no configuration — just paste and go.

  1. 1

    Paste your feedback

    Head to mimir.build and paste any customer feedback, interview transcript, or support thread into the text area on the landing page.
  2. 2

    Hit Analyze

    Click the Analyze button. Mimir saves your text locally and redirects you to sign in.
  3. 3

    Sign in with Google

    One-click Google OAuth. No passwords, no email verification. Your pasted content is waiting on the other side.
  4. 4

    See your workspace

    Mimir creates a workspace, processes your content, and takes you straight into the three-panel experience — chat on the left, insights and recommendations on the right. You'll see a progress card tracking each stage as Mimir works through your sources.

Creating a workspace

Projects are how you organize your workspace in Mimir. Each project has its own sources, insights, recommendations, and chat history. You might create one project per product, per initiative, or per research question — whatever makes sense for how you think about your work.

From the sidebar

Click the + button in the sidebar to create a new project. Mimir starts a short onboarding conversation in the chat panel to learn about your product, users, primary metric, and the question you're trying to answer. This context helps Mimir surface what matters most.

From paste-and-go

If you land via the paste-and-go flow, Mimir creates the project automatically and skips the onboarding conversation — your pasted content is the context.

You can always teach Mimir more about your business through natural conversation in the chat panel. The onboarding conversation is just the quickest way to get started.

Adding sources to your workspace

Sources are the raw inputs you add to your workspace — customer interviews, survey responses, support tickets, product analytics screenshots, meeting notes, or anything else that captures how people experience your product.

Supported formats

.txt, .csv, .pdf for text documents, and .png, .jpg, .webp for screenshots and images. Mimir uses Claude vision to extract text from images, so screenshots of dashboards, support tickets, or whiteboard notes all work.

How to upload

There are three ways to add sources. Drag and drop files onto the Sources page inbox. Click the upload area to browse your filesystem. Or attach files directly in the chat panel using the paperclip icon — Mimir processes them as sources automatically.

Paste text directly

Don't have a file? Paste raw text directly into the inbox. This is especially handy for copying in an email chain, or quick notes from a customer call.

What happens after you add sources

Once you trigger analysis (either automatically from paste-and-go or manually with the /analyze command), Mimir reads your sources, finds patterns, and surfaces what matters. You'll see a progress card tracking three stages.

  1. 1

    Reading sources

    Mimir reads each source individually, pulling out structured signals — pain points, feature requests, observations, metrics, and notable quotes.
  2. 2

    Finding patterns

    All signals are clustered into themes, ranked by severity and frequency. Themes connect related signals across your sources — so a pain point from an interview and a feature request from a survey might surface as evidence for the same theme.
  3. 3

    Surfacing what matters

    Mimir generates a ranked list of recommendations grounded in your themes and business context. Each recommendation includes a rationale explaining why it deserves attention and which themes it addresses.

You don't have to wait for the full process. As each stage completes, results appear in the workspace panel. A toast notification lets you know when everything is ready.

How long does it take

Speed depends on how many sources you're working with and how large they are.

Single source

About 15–20 seconds end-to-end. You'll have insights and recommendations before you finish reading the progress card.

5–10 sources

Under a minute. Sources are read in parallel (up to 10 at once), so adding more doesn't increase time linearly.

15+ sources

About 1–2 minutes. The pattern-finding step does more work to cluster a larger number of signals, but parallel reading keeps the overall time manageable.

The more sources you add over time, the more patterns Mimir finds. Your workspace gets smarter as you feed it more context. Learn more about how Mimir learns.