Coframe Raises $25M Series A for AI Website Experiments

The approval still belongs to the customer. On Coframe's product website, that commitment sits inside an ambitious proposition for growth teams, with AI models and agents doing more of the work needed to turn an idea into a live experiment. A brand can accelerate production while keeping a person responsible for what visitors actually see.

Coframe announced a $25M Series A on November 9, 2026, co-led by Khosla Ventures and Hanabi Capital, bringing disclosed total funding to $35M. The San Francisco company develops AI-supported experimentation for websites, landing pages and advertising. Antler participated, alongside individual leaders affiliated with OpenAI, Anthropic and Google DeepMind.

Founder and CEO Josh Payne and co-founder Pavlo Razumovskyi, VP of Engineering, are pursuing a market in which a growth team can buy more of the work surrounding an experiment. That includes deciding what to try, producing the variation, checking it and interpreting the result. The financing gives that operating proposition more backing as Coframe expands its ambition across digital experiences.

Coframe is selling the capacity to keep testing

A website improvement has to survive several different kinds of judgment before it can produce a commercial answer. A designer considers the experience, an engineer makes the change work, and a growth team decides what result would justify keeping it. Those responsibilities can turn a small proposed change into a substantial claim on several people's time.

Coframe's disclosed workflow connects those activities inside one program. Its software proposes and ranks experiments, creates variations, writes code, performs automated QA and analyzes results. The company also describes human experts working alongside its AI agents, with the buyer able to direct the work and approve releases. That gives the product a practical relationship with the team already accountable for the website.

The commercial appeal follows from the work an untested idea leaves unfinished. A marketing team can have a plausible explanation for a page's performance and still lack the capacity to examine it. Faster production allows more of those explanations to encounter customers, where a response can replace an internal argument. The value depends on the quality of the experiments and the decisions taken from them.

For a buyer, this changes the discussion around implementation. An existing website and testing tools can remain part of the program, with Coframe describing integration through a JavaScript snippet. The conversation can therefore begin with the growth objective, release controls and measurement process. Teams still need a shared definition of success when the software makes it easier to produce another candidate.

The Series A brings financing and category experience together

The $25M round and $35M total describe different financing measures. The latest announcement gives Coframe a larger cumulative capital base. The technology-company affiliations identify individual participants in the financing, alongside the named venture firms.

Khosla Ventures and Hanabi Capital are backing a company founded in 2023. Antler's investment account emphasizes the expense of assembling the specialists needed for repeated experiments and the potential to connect execution with measurement. That account explains the participating investor's thesis, which remains an interested assessment of the business.

Coframe has also acquired businesses founded by two early Optimizely executives. Bo Mohazzabi is now VP of Go-to-Market, and Steven Ebin is General Manager of Revenue Optimization. Their presence gives the company experience with the category's commercial work as Payne and Razumovskyi develop the AI-supported offering. A customer buying an ongoing experimentation program has people, implementation and commercial support to evaluate alongside model capability.

That combination matters because the work continues after the first variation launches. Somebody has to interpret the result, understand how it relates to the goal and decide what deserves another attempt. An enterprise growth team buying Coframe is also choosing how much of that continuing responsibility to place with a supplier. The financing supports a business whose usefulness has to recur across those decisions.

Customer evidence belongs beside the decision it informs

Coframe reports a 410% increase in Replit's enterprise conversions, a result Antler also cites. The figure gives a prospective customer an example to investigate with its own audience, funnel and measurement criteria in mind. Company and investor accounts establish the attribution, while a buyer's decision still depends on how experiments would operate inside that buyer's business.

Customer approval makes that examination more concrete. A team can ask which variations require review, how automated QA is used and how the program handles the objectives that matter commercially. Revenue, qualified demand and other conversion goals place different demands on an experiment. Faster execution becomes more useful when the people commissioning the work can explain what a successful result would change.

Coframe's broader ambition includes digital experiences beyond the website and advertising work described today. Its release places apps and email on the roadmap, and its careers page invites people to help build adaptive interfaces. Those plans extend the same operating question into more places where a company meets a customer.

For the growth leader, the next handoff is already familiar. A proposed experience needs an owner, permission to reach visitors and an answer that can guide the next decision. Coframe is taking on more of the production between those points, with its new investors financing a company that wants to stay involved as the next experiment takes shape.