Say goodbye to traditional A/B testing. Tailor your app's UI to each user and gain an up to 2–3× target metric uplift with the power of Reinforcement Learning.
Ready for the future of app development? Request a demo at flowrl.ai
Heyy everyone, thanks for checking out our thing!
Me and @maharishiva are proud to share flowRL — a UI personalization service that we've been thinking about making for years, and now, finally, got the courage and the necessary knowledge and technology to do so!
Personally, we've struggled with limitations of traditional AB testing — both as people involved in building products, and as users of those products. AB testing is a pretty crude method for improving product performance. It is based on primitive statistical techniques from more than a 100 years ago. It focuses on short-term metrics, requires a long and involved iterative process with multiple stages, stakeholders and other complexities. And in the end, it can only enable an experience that is the same for every user, neglecting the need for personal touch.
We believe in the future of technology where products address individual needs, evolving beyond the one-size-fits-all era.
Leveraging the power of Transformers and Reinforcement Learning, we're taking cues from groundbreaking models like AlphaZero and GPT. We're excited to bring this next level of personalization to user interfaces. Dive into a tailored digital experience with us!
@number16 Can't get enough of your incredible launch! 🚀 It's exactly what our team has been searching for. Best of luck as your project continues to skyrocket. Just became your newest Twitter follower!
I followed some A/B testing app on Twitter and it was a simple imported JavaScript file that allowed you to modify your UI at the percentage that you want. Is this similar to that? Is it that simple or does this require more effort?
@kingromstar not at all!
Well, it IS similar in terms of ease of integration (just a couple of lines of code from our SDK to set up events and UI Variants), but under the hood it's much more complex.
We use Reinforcement Learning to predict the best UI Variants (of the ones you set up in code) for each user, so there is little to none manual input required — our model automatically adapts the interface to drive metrics of your choice!
@kingromstar
Thank you for your question! Traditional A/B testing tools, like the one you mentioned, typically estimate which version performs better on *average* for all users. In contrast, flowRL uses Reinforcement Learning to go beyond averages and tailor experiences for each individual.
This often leads to 2-3x higher uplifts compared to standard A/B tests. It's a deeper approach, but the personalized results are transformative!
Congratulations on the launch of flowRL! This sounds like a game-changing approach to enhancing user experiences. Can you share any specific examples of how flowRL has already improved a digital experience for users?
@candrefs
Hey André, thank you for the kind words!
Indeed, we've been running some early alpha tests, and the results have been promising! One specific example is with a food delivery app to prioritize different layouts with entry points on the homescreen.
With AB testing, we saw a +3.1% increase in conversion rate (CR) for one of the variance. However, with flowRL for personalized home screen experiences, the CR jumped to +6.7%. We're pretty early but super excited to explore more applications and higher levels of personalization to even more products.
Thanks for your interest!
@asaf_pedro
Thank you for the enthusiasm! If you have an application and you'd like to test with us, please book a demo at https://flowrl.ai/ We're excited to share more with you and see how flowRL can enhance your users experience!
Congratulations on the launch of Flowrl.ai! but how it compares to traditional A/B testing. Can you provide some specific examples of how Flowrl.ai has helped to improve target metrics for your customers?
@sarvpriy_arya hey, thanks, sorry for the late response! While an A/B test would arbitrarily separate the user base to find the best "overall" performing solution, flowRL allows companies to find specific UI combinations that work best for each individual user! It allows for
1) Better performance — our models already result in target metric uplift that is 2-3x better than from a "one-fits-all" A/B testing approach
2) Run multiple experiments in parallel, all conflicts are resolved automatically
3) Continuous learning — while A/B testing resolves to a very iterative process, our models don't stop learning and improving. Ever. The more data — the greater the prediction accuracy, and therefore performance, even with the same UI variants
4) Target high-level metrics such as revenue or LTV instead of intermediary ones, like conversion rate on a specific screen
Looking forward to giving this a try!
Funny enough, I've been on the lookout for an exciting A/B Testing platform ever since Google Optimise closed it's doors... then you guys have popped up 👍
@sai_sharan_tangeda either that, or you can connect to your events database. Our models require event data for training and predicting the best UI combinations, but we don't require any specific strings in events, such as "purchase", or "button_press", etc.
Thus, the data can be completely anonymized to hide not only user data, but also your specific event architecture. That's the beauty of blackbox algorithms!
p/general
p/ama
p/introduce-yourself
p/self-promotion
p/thoughts-5
p/zencoder
p/producthunt
p/wisprflow
p/kraftful
p/screaming-frog-seo-spider
p/xnapper
p/google-chrome
p/fine
p/openai-api
p/bolt-new
p/elevenlabs
Replies
flowRL
Go Mail Merge
flowRL
flowRL
flowRL
PrevessAI App
flowRL
flowRL
flowRL
flowRL
OkFeedback
flowRL
flowRL
flowRL
Glory Morning
flowRL
flowRL
SigmaChat AI - Practice Negotiation
flowRL
flowRL
Meshy