Home/Use cases

Which AI agents to use for what

Understand users

Decide what to build

Design and prototype

Build and ship

Every setup in one line

What's the best AI setup to synthesize user interviews into themes?

Bring the transcripts and the rest of your feedback into Dovetail, which turns them into insights that cite their sources. Then use Claude to shape the themes into a readout people will actually read, and log what matters in Linear.

What's the best AI setup to run user interviews at scale with an AI moderator?

Outset writes the interview guide from your goals and runs hundreds of AI-moderated interviews at once. Pick Listen Labs instead when you need many languages. Either way, keep the results in Dovetail so themes add up over time.

What's the best AI setup to test a prototype with real users?

Build the prototype in Figma Make, then run an unmoderated test in Maze, which recruits people from its own panel and writes the report. You get real reactions without scheduling a single call.

What's the best AI setup to turn customer feedback into a spec?

Samepage Bets pulls feedback from calls, tickets and chat into one place and ranks it into scored bets. Take the winning bet into ChatPRD to write and score the PRD, then hand the work out in Linear.

What's the best AI setup to write a PRD from meeting notes?

Use ChatPRD. Paste in your meeting notes and it drafts the PRD, scores it, and can open it in v0, Lovable or Cursor when you're ready to build. Keep the final spec where your team works with Notion Agent.

What's the best AI setup to run a competitive analysis?

Send ChatGPT agent to browse competitor sites and build the comparison, ask Claude to pull out the gaps and opportunities, and save the brief in Notion with Notion Agent. Check every claim against the sources before you share it.

What's the best AI setup to turn a PRD into a clickable prototype?

Draft or tighten the PRD in ChatPRD, open it in v0 to build a working web prototype, then put it in front of real users with a Maze test. You end up with a clickable app and real feedback on it, often in a single day.

What's the best AI setup to generate UI that uses your own design system?

Start in Figma with the Figma agent, which builds screens from your real library components, tokens and variables. When you want to see it as code, move into Wonder, where the design stays as code and a coding agent builds the version you pick.

What's the best AI setup to turn a whiteboard sketch into a prototype?

Photograph the sketch and let Stitch turn it into a digital design, paste that into Figma to clean it up, then make it clickable with Figma Make. If you're not a designer, Uizard is the gentler start.

What's the best AI setup to prototype a new feature inside your real product?

Use Wonder. It's a design canvas on your actual product where the design stays as code, and it runs a coding agent like Codex to build what you design. If you just want a quick build from a GitHub repo, Bolt can start from the repo and keep your design system.

What's the best AI setup to turn a Figma design into production code?

If the work has to land in your real codebase, give the design to a coding agent that works in that repo. Wonder keeps the design and the code side by side and runs Claude Code for you, and GitHub Copilot reviews the pull request. For a brand new app, Bolt can start a build straight from your Figma file.

What's the best AI setup to get design QA fixes shipped?

Log each fix in Linear and delegate it straight to a coding agent like Cursor or Codex, then review what comes back with GitHub Copilot code review. Small UI fixes stop waiting for a sprint.

What's the best AI setup to build a working app without writing code?

Lovable is the default. Get the idea down as a spec in ChatPRD, open it in Lovable, and it builds and deploys a full app with hosting, sign-in and payments handled. Pick Replit Agent if you want it to test the app in a browser and fix its own bugs as it goes.

What's the best AI setup to automate status updates and stakeholder reports?

Use the agent that lives where your work does. Notion Agent can run on a schedule to post updates and route tasks, and Claude can draft the part that needs judgment. Teams on Jira and Confluence should use Rovo, and research teams can let Dovetail's agents send recurring stakeholder updates.

New agents, in your inbox.

Get the newsletter