Today we're introducing Wave. A proactive product agent that helps teams build self-improving products. Every product team runs the same loop. Build, ship, use, learn. AI has made building and shipping extraordinarily fast. Understanding usage and learning still happen by hand. Wave runs the whole loop: - Analyzes your... Amplitude data for signals across analytics, feedback, session replays, error logs, agent traces, and experiment results - Surfaces opportunities in the form of full product specs that you or your agents can approve and ship - Tracks the experiments and outcomes of what you've built so you can start the loop over again Your product learns what works and gets smarter every cycle:show more

one fun story on a little insignificant feature identified and shipped by wave 🌊 inspired by listening to @lennysan's podcast with @TheAmolAvasare where he shared anthropic had built an internal experimentation system called CASH to identify opportunities for growth experiments, we pointed wave @ our corp site the following week to look for mini-optimizations for fun. our agent identified that the conversion rates to sign-up on our blogs could be improved, figured out from session replays and our codebase that we didn't have follow-up actions there, then hypothesized about an experiment for just adding a quick CTA to drive sign-ups. we punted the opportunity to a @cursor_ai cloud agent which one-shotted adding the CTA to the blog, and surprisingly we're driving 100's of new click-throughs for getting started on amplitude. super little feature but just pure fun to have an agent watching my products and flows behind my back 😂

did somebody say "loops"? so excited to be self-improving loops to your product with wave 🌊

Should we switch to loops as the product name? Wave is too nerdy according to @gr8ful_nate

So excited Amplitude is getting this out! Wave is a paradigm shift in bringing experimentation-based reinforcement learning on products to the masses. And it's easy to use...

If you're an AI builder, Wave is for you! And if you're interested in joining our amazing team working on this, reach out! We're hiring:

congrats on the launch, Steven!

Been so cool to see little improvements proposed and built by AI every week Here's an example problem -> solution wireframe I received this morning in Slack

that's really cool. loops on loops.

Wow. Will never forget when you first told me about this last year. Finally live!

Thanks Turner!!

The holy grail of product analytics was never the dashboard. It was closing the loop from signal to shipped. Super excited for this!

I remember when you and I were talking about this 10 years ago. Crazy it is now here!

hello Wave! Love this direction for Amplitude - honestly product data is the only way to push back on the deluge of slop code and slop features inundating all of us right now. It's great agentic coding is "solved" but we really really need agentic "product"

ride the WAVE baby $AMPL

Have you talked to our IR team? You guys should connect

We've been chasing the Self Improving Products idea for over ten years. The models weren't good enough, nobody had behavioral data at this scale, and no one trusted software to change itself. That's all different now. We can finally close the loop.

I love seeing the vision of self-improving products come to life!

The shift is subtle but massive. Product teams do not just learn from the data anymore. The product starts learning with them.

This has been my all-time favorite surface to show customers and partners. It is crazy how good it is today.

finally realizing the dream of a self improving product ❤️

when our dreams combine the sky is the limit

This is huge 👏

Crazy!

Really cool! Congrats Spenser and team!

Thanks Tomer!! I'd be excited for Gusto to try it out

the future is now!

The biggest bottleneck in product development is increasingly shifting from building features to understanding what actually works after launch.

The build ship learn loop is fascinating when applied to education products. Every student interaction on Docuee teaches us something. Where they get stuck in their writing. Which AI suggestions they accept or reject. Which stages cause the most supervisor feedback. Where projects get abandoned. But here is what makes edtech different from most products. The student cannot opt out of the learning loop. They have a deadline. A supervisor. A defence to prepare for. That constraint makes the feedback signal unusually clean. Students do not use Docuee casually. They use it under genuine pressure. And products used under genuine pressure reveal exactly what works and what does not faster than any A/B test ever could. The best product research is a student at 2am who has no choice but to make it work.

the learn part of the loop has always been the bottleneck. every PM I know is drowning in that step. curious how wave handles the signal/noise

Great minds yada yada.

incredible you came to the same point of view!

Bet many do. Seems like a reasonable way to remove friction in terms of software maintenance.

Or if they don’t yet, they’ll figure it out soon when agents truly agent at that level of output quality.

*reach

I’d want it to catch the quiet product regressions first.

We do! Want to use it?

The category just got real. We've been building the same loop at but starting from full PM context (calls, tickets, specs), not just the event stream. Congrats to the team 🌊

Self improving products, woah! Congratulation on Wave👏 Looking forward to try it's power very soon

awesome.

build and ship got 100x faster. the loop was always bottlenecked at "learn". this closes it

this looks so amazing! this feels like how you described the vision a decade ago!

Congratulations on great 2Q results. Statsig acquisition is proving to be an incredible decision. You are truly repositioning the company for the Ai future

loops. it's loops.

@ycombinator Congrats on the launch 👏

build product, ship to agents, agents use product and give feedback, iterate off a agents, this seems to be next generation of product loops

That’ll shape how we works in closing the product iteration feedback loop. Exciting!

Build and ship got cheap. Understanding stayed expensive. Knowing which signal is real and which is noise is the part that still needs a human, and it's quietly becoming the whole job