Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.replit.com/llms.txt

Use this file to discover all available pages before exploring further.

⏰ Estimated time: 15 minutes A prototype is the smallest version of a product idea that someone can try and react to. You will use Pace as the example: a premium running tracker that gives runners a polished overview of their training, recovery, and progress.
The Pace prototype published at a public URL — a premium running tracker dashboard with weekly distance, recent runs, upcoming workouts, and achievements

​
What you’ll learn

You will learn how to:
  • Turn a broad product idea into one testable flow.
  • Use Plan mode to scope a prototype before building.
  • Review whether Agent’s plan is small enough to test.
  • Use Canvas annotations and collaboration to collect feedback in context.
  • Explore two prototype directions in parallel when the product question is unclear.
1

Write the prompt

The prompt above describes Pace — a personal running tracker with a polished home dashboard, an Add Run flow, stats, a placeholder plan view, and realistic San Francisco sample data. Use Copy the prompt to paste it into Replit yourself, or Build on Replit to open Replit with the prompt already attached.The prompt is intentionally specific about the visual direction, the screens, what data to mock, and what’s a placeholder — that’s what gives Agent a clear target to evaluate against.
2

Use Plan mode (optional)

If you want Agent to think through the product question before writing any code, toggle Plan next to the prompt input. Plan mode produces a structured plan you can review before approving the build — useful when the direction isn’t fully decided yet.
The Replit prompt composer with the Pace prototype brief typed in and the Plan checkbox highlighted, ready to scope the build before Agent starts
Review the plan: target user, main flow, what should be real, what can be mocked, and a test checklist. If the plan adds too many features, ask Agent to reduce it.
3

Review what Agent built

When Agent finishes, open Preview and use Pace like a first-time visitor.Check that you can:
  • See the Alex Morgan dashboard populated with realistic sample runs.
  • Log a new run from the floating + or the hero Start a Run button.
  • See the run appear in Recent Runs and the This Week stats update.
  • Browse the Runs, Stats, Plan, and Profile tabs.
  • Resize Preview to a phone-sized width to check the mobile layout.
If something feels off, describe the symptom to Agent and ask for the smallest fix.
The Pace prototype running in Preview — the workspace shows Agent's chat on the left and the populated Pace dashboard on the right, the kind of surface a first-time visitor would interact with
4

Publish and ask for feedback

Publish when the main flow works.
The Project Editor with both publish entry points visible: the inline Publish card in the Agent chat with a subdomain field and the Publishing tile in the Tools and files panel
Once published, open the public URL in a new browser tab to make sure the experience works for first-time visitors before you share it.
The published Pace prototype open in a new browser tab, ready to share with reviewers for focused feedback
Share the URL with one focused task:
Use Pace as a runner trying to track a week of training. Log one run, browse the dashboard, and tell me what feels confusing or motivating.
Good feedback should help you decide what to build next. Avoid asking, “What do you think?” unless you want vague answers.
5

Collaborate with your team

Product prototypes get better when teammates can review the same artifact and give feedback in context.Invite collaborators based on how closely they need to work with you:
  • Invite someone to the project when they need to review or help with this prototype only. See Invite teammates.
  • Invite someone to a workspace when they are part of an ongoing team and need access across projects. See Team workspaces.
The Invite dialog open in the workspace top bar with the username-or-email input highlighted, ready to invite a teammate to the project
Then, use Canvas when the feedback is visual. Add notes directly on the prototype, annotate the area you want changed, and ask Agent to use those notes when it updates the app.
The Pace frame on the Canvas with a rectangle, arrow, and text label pointing at a 'Log in with Strava button' spot next to the Log Run button — the kind of contextual note collaborators can leave for Agent
For PMs, this means the prototype works as a shared artifact: designers can annotate flows, engineers can inspect feasibility, and stakeholders can try the published version.

​
🎉 Congratulations

You’ve turned a product idea into a focused, shareable prototype — a working Pace dashboard that teammates and customers can try and react to. The hard part wasn’t this prototype. It was learning how to reduce a broad idea to one testable flow, collect feedback in context, and decide what’s worth building next.

​
Next steps

Use feedback to choose one next step:

Add a database

Save run history so it survives refresh.

Add login

Give each runner an account with their own progress.

Build a mobile app

Take Pace to phone-first tracking on iOS or Android.

Create a dashboard from data

Turn run history into progress insights.

​
Build in parallel

When the product question is about direction, not just implementation, exploring multiple ideas at once helps you compare options instead of committing too early. Replit Agent supports parallel tasks — you can ask Agent to work on more than one prototype, feature, or experiment at the same time and review them side by side.
Thread view showing multiple Agent tasks running in parallel with status indicators
For how tasks queue, run, and finish, see Task system. When you are not sure which direction is better, ask Agent to create two parallel tasks. For Pace, you could compare:
  • Motivation-first prototype: streaks, achievements, encouraging recommendations.
  • Analytics-first prototype: deeper stats, week-over-week trends, personal records.
After both tasks finish, compare them with the same feedback question:
Which version makes you more likely to come back tomorrow, and why?

​
Need further help?

  • The prototype has too many features: ask Agent to reduce it to one target user and one flow.
  • The audience is unclear: ask Agent to rewrite the welcome section for one target user.
  • Mocked data hides risk: ask Agent to list what is real and what is mocked.
  • Feedback is vague: give reviewers one task and one question.
  • Publishing fails: use Troubleshoot publishing.