Launching soon

Tulir

Maze tests prototypes. Hotjar replays behavior. Sprig sends surveys. Tulir is different: it asks a real user to attempt one specific task inside your live product, records the attempt in context, and returns a one-page friction audit you can act on Monday.

Get a founding audit — $499  → All products
Made with
  • TypeScript
  • Next.js
  • Hono
  • PostgreSQL
  • LiveKit
  • Claude AI
What it does

  1. In-product task testing

    Not a prototype link. Not a survey. Real users completing one specific task inside your live product (onboarding, activation, invite flow, billing), with full screen and camera recording.

  2. 48-hour friction audit

    Top 3–5 friction points ranked by severity. A replay clip jumping to the exact moment each one broke. Verbatim quotes when voice was enabled. Suggested fixes from an AI that watched all five sessions.

  3. One SDK, zero noise

    A single script tag on your head. The SDK only records participants you send a Tulir link, never your normal traffic. No GDPR landmines, no session replay for everyone.

  4. AI analysis built on full context

    The AI reads your task definition, watches all five sessions, and writes findings against your specific goal, not generic UX heuristics. It knows what you were trying to test.

  5. MCP-native for AI builders

    If you're building with Claude Code, Cursor, or Windsurf, Tulir has an MCP server. Your AI coding agent can trigger a friction audit and read the results directly in the conversation.

  6. Private shareable report

    One link per audit. Send it to your team, PM, or investor. No account required to read it. Built to survive a Slack DM, not a tool that requires onboarding before it's useful.

Tulir isn’t public yet. Drop a note via the contact link below if you’d like early access or a heads-up when it ships.