Coframe raises $25M Series A to make websites self-improving and agent-ready, citing 70% conversion lifts

Coframe raises $25M Series A

Coframe has closed a $25M Series A. The company's pitch is that websites, ads, apps, and emails should improve themselves continuously rather than requiring constant input from engineering and design teams. CEO Josh Payne describes the status quo as costing hundreds of billions of dollars in lost conversions annually.

The product enters via a JavaScript snippet. Payne says one Fortune 500 customer went from signed contract to live deployment in three days, then saw a 70% lift in conversion within a month. He cites similar results elsewhere: 158% at Dropbox, 410% at Replit, 60% at Cash App.

“High level idea here with Co Frame is that the Internet is dead... we see a world where the internet itself can have its own sense of life, where it can self improve and become this kind of continuously learning system. With that customer, we are able to see a 70% lift in conversion in less than a month. By 2029, the majority of traffic in the US for commerce is gonna end up being agentic personal agents.”

Pricing and model

Coframe currently charges a flat annual rate, targeting at least 10x ROI as the internal threshold before proposing a contract. Payne acknowledges performance-based pricing is the more logical model given those numbers but says the company is keeping it simple for now. The obvious tension is that better results justify higher fees, but rising AI compute costs eat into margins until model prices fall.

The agent-readiness bet

The longer play is infrastructure for agentic traffic. Coframe lets any site it powers serve an agent-optimized version by appending .agent to the URL, using protocols like WebMCP. Payne says this isn't driving customer budget decisions today and won't be for at least another year, but he predicts agentic personal agents will account for the majority of US e-commerce traffic by 2029. The go-to-market approach is to sell on today's conversion problem and position agent-readiness as the included upgrade.

Nat Friedman led Coframe's seed round. The company is now 45 people, though Payne says revenue per employee is the number he cares about, not headcount. His stated end state is a company structured like an AI lab, where the entire team focuses on improving a proprietary transformer trained on the user-interaction data flowing through the platform.

The $25M gives Coframe runway to pursue that model-training flywheel. The conversion numbers, if they hold at scale, make the pricing conversation straightforward. The harder question is whether the agent-readiness layer becomes a genuine moat or a feature any CDN or CMS can replicate once the protocol standards settle.