Runner AI Wants to Build Your Store and Then Run It. The Second Part Is the Actual Bet.

The Macro: Your Store Is Live. Now What?

Shopify reported Q4 2024 revenue of USD 8.88 billion, up 26% year over year. The website builder software market sits somewhere between USD 1.9 billion and USD 3.9 billion depending on which research firm you trust, and analysts broadly agree it’s growing through the end of the decade. Point being: there is a lot of money moving through the act of putting products on the internet.

And yet the conversion rate for the average e-commerce site in 2025 is 2.91%, according to Dynamic Yield data cited by Forbes. Two and a half percent of visitors buy something. The rest leave.

That gap is where everyone is fighting right now.

The incumbent pitch, from Shopify to Squarespace to Wix, is essentially: here are tools, good luck. They’ve gotten dramatically better at the build side. Templates are smarter, storefronts are prettier, the barrier to launch is basically zero for someone with a product and a weekend. But optimization, the ongoing work of figuring out why visitors leave and fixing it, still falls on the merchant. You hire a CRO consultant, you run A/B tests manually, you watch heatmaps and make educated guesses.

A newer wave of tools is targeting that gap directly. FERMÀT builds shoppable landing pages designed to lift conversion. Cuped.ai positions around experimentation infrastructure. The thesis across all of them is roughly the same: the build is a commodity now, and the value is in what happens after launch.

Runner AI is making the same bet. But it’s trying to own the build step too, which makes the story more interesting and the execution harder.

The Micro: Prompt to Store, Then Leave It Running

The core idea is straightforward. You describe what you want, upload some images, and Runner AI generates an e-commerce storefront. Winter clearance sale, home decor boutique, personal brand site. The product page shows prompt examples exactly like that, keeping expectations grounded and specific rather than abstract.

That part is table stakes in 2025. The vibe coding wave has produced a lot of tools that let non-technical people build functional sites through natural language. Anima has been chasing the gap between design and shipped code for a while now, and the broader vibe coding moment has gotten strange enough that someone built a gamepad for it. Building with a prompt is not a differentiator anymore.

What Runner AI is actually selling is the layer that runs after you publish. The platform claims to continuously run experiments in the background, automatically. Not “here are A/B testing tools you can use.” Automatically. The pitch is that the system watches visitor behavior, identifies friction, tests solutions, and adjusts the store without you scheduling a sprint.

That is a meaningfully different product promise than anything in the Squarespace category. It’s also a harder one to verify from the outside.

The Shopify remix option on the homepage is notable. Runner AI isn’t asking you to abandon your existing setup entirely. It hints at a migration or overlay path, which lowers the activation cost considerably if it works as described.

It got solid traction when it launched, including a notable mention framing it as a self-optimizing e-commerce engine. The framing landed.

For e-commerce merchants who already know that product photography is half the conversion battle, the appeal is clear. Less time managing the store mechanics, more time on the product itself.

The Verdict

The build-with-a-prompt piece I believe. That works now. Multiple tools have demonstrated it.

The autonomous optimization piece I want to see receipts on. “Continuously runs experiments automatically” is either a genuinely differentiated capability or a polished way of describing something much more limited under the hood. I can’t tell from the outside, and that ambiguity is the whole product.

At 30 days, the question is whether early users are actually seeing the optimization loop do anything measurable. A conversion rate that moves from 2.5% to 3.5% without manual intervention is a compelling case study. Noise dressed up as signal is a churn driver.

At 60 days, the Shopify integration story matters a lot. If merchants can layer Runner AI onto an existing store rather than rebuild from scratch, the addressable pool gets much larger. If it’s a migration play, the sales cycle gets much harder.

At 90 days, I’d want to know if the autonomous part requires ongoing prompt input or genuinely runs without the merchant touching it. That distinction is the entire thesis.

I’m interested in this one. Not convinced, but interested. The bet they’re making is correct. Whether they can actually execute the optimization layer is a different question entirely.