Hi, it's Rohan and Peter from Ripple.
Things were changing so fast in our product and in the world, that it became hard to commit to a website. Instead we thought we'd just outline how we're thinking about email, where it's headed and what we're building.
We've always believed that AI should enable email programs to perform better than they do today, at a significantly lower labor cost (compared to the large teams that are required today to run these programs).
We believe a truly AI native email program will enable all of the following:
- Personalized messaging to every person, at the right time.
- Continuous a/b testing on content, send times, subject lines, frequency etc.
- Automatic notifications for things happening in the business (new arrivals) or in the world (weather changes)
- A continuous feedback loop for the AI to learn from what’s working and self improve
To Do This, Design Had To Be Automated
The bottleneck today for all of this is that producing emails is just too time consuming and expensive.
For the last few years, we've taken many cracks at trying to automate design, but candidly none of them worked.
In our latest iteration, we believe we're close.
All the emails below were 100% generated by AI from just a prompt, with no editing, and no templates.
Dynamic Customer Journeys Vs Static Flows
Automated design opens up a whole host of interesting possibilities, but one we were particularly excited about was this idea of dynamic customer journeys vs static flows (like exist today in Klaviyo).
Static flows are not only a pain to set up and keep up to date, they are also not optimized for conversion.
Instead of every person who adds something to their cart going through a static abandon flow, every person in the brand's email list should go through their own individual journey, orchestrated by a centralized “brain”, that uses all the information we have about them in that moment, and can also get smarter based on data on what’s working.
We’ve built this into Ripple and are testing it with a handful of customers.
Below is an actual customer journey orchestrated by AI. You can see how it seamlessly adapts from an abandon series to a post purchase to a cross-sell.
At every moment, it uses all the information we have about the customer, including their purchase history, their browsing patterns, whether their order has been delivered, and their engagement levels. It also has information about what’s happening with the brand, like live promos and new arrivals.
Using all of that, it can plan a bespoke journey just for that customer.
Hannah had no purchase history, but she had a high-intent cart. This was sent 10 minutes after abandonment to capture peak recall, using the exact products she had already selected and Offe's natural urgency around fast-moving inventory.
Once her 8-item first order landed, the journey switched instantly into Order Confidence. The goal here was to validate her taste, make the purchase feel intentional, and use the immediate post-purchase window to reinforce trust and excitement.
A day later, the system moved from reassurance to taste validation. Because Hannah bought multiple jewelry pieces, this email framed her order as proof of judgment, pushing Offe from discount retailer toward trusted curator in her mind.
Hannah came back and viewed the Black Sheep Coin Pendant Necklace three times, just after her first jewelry order. That signal was strong enough to trigger a targeted recovery email that tied the necklace back to her recent haul while she was still actively browsing.
This step shifted from recovery into utility. By showing Hannah how to style pieces she already bought, the journey made Offe feel less transactional and more editorial, while keeping engagement high without repeating the prior pitch.
The delivery-arrived moment created a natural cross-sell window. Instead of pushing something random, the system used that post-delivery excitement to gently reintroduce the exact necklace Hannah had already viewed repeatedly, making the recommendation feel timely, intelligent, and earned.
After several jewelry-focused touches, the system widened the frame. Hannah had already shown interest in Living and Dresses, so this email expanded Offe beyond accessories and pushed toward a broader browsing habit instead of more of the same.
Although we’ve only tested this with a few customers so far, results have been extremely encouraging. That brand we shared the example from, saw a 9x lift in flow revenue within the first week of moving from static flows to journeys. More on that here.
So What’s Next
There’s a lot to build. We believe we’re just scratching the surface on what an AI native email program would look like and how it can perform.
We’re grateful to be working with a set of forward thinking design partners as we tinker, experiment and build.
If that sounds like something you’d be interested in, please leave your email below and we’ll reach out if it’s a good fit.