Coframe, a startup that builds AI to rewrite websites for each visitor, said Monday, Nov. 9, that it raised a $25 million Series A co-led by Khosla Ventures and Hanabi Capital. The round brings its total funding to $35 million, according to the company's announcement.
The pitch is a direct challenge to A/B testing, the long-standing practice of showing two versions of a page to different audiences and keeping the winner. Coframe wants models that change a site on their own. Its funding post describes customer experiences as multimodal reinforcement-learning environments, fed by a dataset the company says grows by hundreds of millions of sessions a month, with models optimized directly for revenue. The post offers little technical detail. It does not disclose model size, architecture or how the systems are evaluated.
What the funding numbers do and don't say
Silicon Report, which covered the round on Nov. 10, reported that Coframe did not disclose a valuation. Antler, another investor, confirmed its participation in a Nov. 9 blog post by partner Chris Walsh but did not say how much it put in.
The cumulative figure deserves a caveat. PitchBook and AlleyWatch both put the earlier seed round at $9.3 million, which, added to the $25 million Series A, comes to $34.3 million. Coframe says $35 million. Rounding or additional earlier funding could explain the roughly $700,000 difference, but neither explanation was verified in the reviewed sources, and no public clarification from the company was located.
Reuters reported the seed round on Oct. 29, 2024, and described Nat Friedman as an established backer through his firm NFDG.
Some coverage got the arithmetic wrong. VentureCapital.com's Nov. 9 headline described a $35 million Series A, which conflicts with Coframe's own split of $25 million for the Series A and $35 million in total.
For scale, Postround reported on Nov. 16 that 38 Series A rounds totaling $2.8 billion were announced the previous week, with a median of $24 million. Coframe's $25 million sits in that dataset, which excludes China, rounds of $8 million or less, and therapeutics companies. In other words, the round is solid but typical.
Customers, claims and the roadmap

Coframe's release names Dropbox, BCG, Block, Backcountry and Sur La Table as customers. These are company-supplied claims, and the extent of each relationship is not described.
The company said apps, email and other digital experiences are next on its roadmap, which means they are not yet products. The Nov. 9 press release does not provide launch dates.
The release also said Coframe is working with OpenAI on ChatGPT ads, tying ad optimization to landing-page optimization. No separate confirmation from OpenAI could be found, so for now the collaboration rests on Coframe's word.
Antler offered a description of the underlying technology. By its account, Coframe measures clickstreams linked to each page variant and carries learning across customers through model gradients rather than raw data. That is an investor's characterization. No technical paper or privacy audit validating the architecture turned up.
Coframe's headline customer result is a 410% conversion lift for Replit. The case study is several months old, and a lift on one page or test does not translate into a company-wide increase in sales or revenue.
The team and the deals

CEO Josh Payne completed undergraduate studies and left Stanford's MS and MBA programs without finishing, according to the release. The company also said it acquired businesses founded by Bo Mohazzabi and Steven Ebin. Coframe announced its HaystacksAI acquisition on March 4, 2026. The target, timing and price of the deal tied to Ebin could not be established.
Where Coframe sits in the market
Antler's investment thesis is that the real bottleneck in experimentation has been how quickly teams can build and ship variants, not the statistics alone. That is the view of an interested investor, not an independent assessment. None of the reviewed sources, including the company release, Antler's investment note, Business Insider's feature and Silicon Report's brief, included a direct competitor comparison, so how Coframe stacks up against rival tools is hard to judge from the outside.
What is clear is the next test. Apps and email are the stated next frontier, and until Coframe offers a timeline, or independent evidence for its technical and customer claims, the story remains largely the company's own.
