Product Agent

The Product Agent installs your app (or a competitor's) on a real device via BrowserStack, navigates through the onboarding flow using computer vision, captures screenshots, and generates a detailed UX analysis report. Think of it as an AI product manager reviewing your first-run experience.

Status: Live in production

What it does

The Product Agent automates the tedious work of manually testing app onboarding:

1. Spins up a real iOS device in BrowserStack (cloud device farm)
2. Installs the app from the App Store using your provided URL
3. Navigates the onboarding flow screen by screen using an AI vision agent that understands UI elements
4. Takes screenshots of every screen encountered
5. Generates a comprehensive report analyzing the user experience, friction points, and UX patterns
6. Saves everything to your Fload account for future reference

The entire process is automated. You just provide an App Store URL and wait 5–15 minutes.

What the report includes

Onboarding flow analysis

• Screen-by-screen breakdown — what the user sees at each step
• Permission requests — which permissions are requested, when, and how they're framed (notifications, location, camera, etc.)
• Value delivery timing — when the user first experiences the core value of the app
• Paywall placement — where and how you ask for payment (if applicable)
• Friction points — steps that might cause user drop-off or confusion

UX assessment

• First impressions — what new users see the moment the app launches
• Clarity — is the value proposition clear?
• Simplicity — is the flow intuitive or overwhelming?
• Trust signals — does the onboarding build confidence?
• Call-to-action effectiveness — are CTAs clear and compelling?

Screenshots

Every screen is captured and included in the report, so you can see exactly what the agent encountered.

How to run it

1. Navigate to Product Agent in the Fload sidebar
2. Enter the App Store URL of the app you want to analyze (yours or a competitor's)
3. Click Run Analysis
4. The job is queued. Typical runtime: 5–15 minutes depending on onboarding length
5. Once complete, the report appears with full screenshots and analysis

Technical details

Architecture

• BrowserStack integration — real iOS devices (not simulators)
• Vision agent — powered by AI vision models (Claude Sonnet 4.5) that understand UI elements
• Automated interaction — the agent clicks buttons, scrolls, fills forms, and handles common flows
• 2FA handling — can handle SMS 2FA via Twilio integration if configured
• Playbook system — successful flows are saved and can be replayed for faster future runs

Supported platforms

• iOS — full support via App Store
• Android — not currently supported (Google Play coming soon)

Limitations

• Public apps only — the app must be live on the App Store (TestFlight builds not supported)
• One run at a time per organization (concurrent runs coming)
• iOS 16+ — older iOS versions may have compatibility issues
• English UI — the vision agent works best with English-language apps (multilingual support improving)

Use cases

Pre-launch competitive analysis

Run the agent on 3–5 competitor apps to understand:

• How they onboard new users
• When they ask for permissions vs. deliver value
• How they present their paywall
• What friction points exist in their flows

Post-update validation

After shipping a new onboarding flow, run the agent to:

• Verify the flow works end-to-end on real devices
• Catch bugs or broken screens before users report them
• Confirm permission prompts appear at the right time

Team alignment

Generate a report and share it with your team so everyone understands:

• What the current new user experience actually is
• Where users might drop off
• What needs improvement

Design review

Use the screenshots as design documentation:

• Show designers what shipped vs. what was mocked
• Identify inconsistencies between platforms
• Document the onboarding flow for future reference

Pricing & usage

Usage is tracked per calendar month. View your usage at Usage in the sidebar.

Common questions

Can I run it on a competitor's app? Yes. You just need a public App Store URL. This is one of the most common use cases.

Does it test TestFlight builds? No. The app must be live on the public App Store because the agent installs from the store, not via TestFlight invites.

What if my app has a signup flow? The agent can handle signup flows if you configure test credentials. Contact support to set this up.

Can I see the agent in action? Not in real-time, but you get the full screenshot history and report after completion.

What if the agent gets stuck? The agent has a 15-minute timeout. If it gets stuck, the run fails and you're not charged. You can retry or contact support.

Roadmap

• Android support — Google Play installs via BrowserStack
• Concurrent runs — run multiple analyses in parallel
• Custom test scenarios — define specific flows to test (e.g., "complete a purchase")
• Scheduled runs — automatically re-test your app weekly or after builds
• Comparison mode — diff two onboarding flows side-by-side