We just raised $2M from Google’s AI fund to build self-improving software. We’re starting by automating paywall testing for mobile apps. 🧵

When I was at Uber, my teams made millions testing new in-app experiences, but saw how long it took to test even basic changes.

1. Come up with ideas (often by looking at other apps), 2. Rank them 3. Have a designer mock them 4. Create a final version 5. Have an engineer implement it 6. Configure the experiment 7. Wait 6 weeks, learn that it did ~ ok ~, then try again with something new

Not to mention half the time something breaks with testing or rollou

But given the speed LLM’s are accelerating at both design and software development, it was hard for me to imagine this flow looking the same in 5 years, or even 2

That’s why we’re building Helium — AI that automatically generates and tests new in-app experience to improve revenue.

We’re starting by focussing on mobile apps — the birthplace of much of modern software experimentation (Zynga, Meta).

Our first product lets growth teams automatically generate new mobile paywalls, then immediately run tests to improve conversion.

Then Helium’s AI leverages hundreds of past datapoints to decide what to test next, generating a custom paywall that implements then next test idea.

We've been building with some of the fastest-growing apps in the world for over a year — and have seen a >40% average increase in revenue during the first since months.

We're currently serving many millions of paywalls a day, and are finally opening a public waitlist!

If someone you know is building a mobile app, and wants to get more out of their paywall, take a look:

We'll also working on some other pretty exciting features dropping soon, so sign up on the website to stay in touch.

Stop moving buttons around. Start making money in your sleep. 💸✌🏼

If you're not learning AI in 2025, you're falling behind. Join 1,000,000+ early adopters reading and learn AI in just 5 minutes a day (for free).