Coframe raises US$25M as AI websites move beyond A/B testing
Coframe has raised US$25M to automate website experiments, but its reported conversion gains need independent validation.
Coframe has raised US$25 million in Series A funding to build websites that generate, test and deploy changes continuously rather than wait for the next manual optimization cycle. The San Francisco company is applying AI agents to a part of marketing still dominated by experimentation queues, design reviews and engineering tickets.
The round, co-led by Khosla Ventures and Hanabi Capital, brings Coframe's disclosed total funding to US$35 million, according to its November 9 announcement. The pitch is that landing pages, ads and eventually other digital experiences can adapt to visitor behavior at a speed a human growth team cannot match. Whether that delivers a sustained advantage over existing optimization tools is less settled.
Key Takeaways
- Coframe has secured a US$25 million Series A to expand its AI-based website experimentation platform.
- The product combines generation, testing and deployment rather than acting only as a copywriting tool.
- Reported conversion and customer-revenue gains are company claims that marketers still need to validate independently.
Table of contents
Jump to each section:
- A US$25 million round backs an always-changing website
- Coframe is automating the work around experimentation
- The performance claims still need independent proof
- What marketing teams should test before handing over the website
A US$25 million round backs an always-changing website
Coframe's new financing follows an earlier seed round and extends an investor bet on automated conversion optimization. Its backers include Khosla Ventures, Hanabi Capital, Nat Friedman and Antler. The valuation was not disclosed. The size of this round matters because the company is trying to expand beyond the traditional testing-tool category into a broader automated web-experience layer.
US$25 million Series A; US$35 million raised overall. Those are Coframe's announced financing figures, also reported by Dealroom on November 9, 2026.
In practical terms, Coframe wants to turn a marketer's hypothesis into live variations with less coordination among analysts, copywriters, designers and developers. That is a potentially attractive proposition for teams with large traffic volumes but a limited capacity to run experiments, especially when they need different content for multiple customer segments.
Founder and CEO Josh Payne describes the ambition in a short line: "We're building an internet that improves itself." It is a useful statement of direction, but it should not be confused with evidence that automated decision-making always improves the customer experience.
Coframe is automating the work around experimentation
The underlying proposition is more substantial than asking a chatbot to rewrite a headline. Coframe says it uses a model trained on user interactions alongside agents assigned roles such as strategy, design, engineering, quality assurance and data analysis. These systems generate alternatives, put them into experiments and use results to guide subsequent changes.
That collapses several stages of a conventional CRO workflow. Instead of preparing a few page versions for a testing calendar, a marketer can potentially manage a larger set of ongoing variations. The central decision shifts from which variant to build next toward which objectives, audience boundaries and experiment safeguards the system should follow.
There is already competition for that position. Established platforms such as Adobe Target, Optimizely and VWO offer experimentation and personalization; newer specialists are trying to automate more of the creative and operational work. Coframe's differentiation claim is not merely better testing software. It is that more of the testing workflow can run with less manual intervention.
The company's November announcement also describes work with OpenAI around ChatGPT ads and landing-page optimization. That is Coframe's account of the relationship, not an independently verified claim of advertiser performance or an indication that OpenAI itself invested in the company.
The performance claims still need independent proof
The most striking numbers in the announcement concern customer results. Coframe says its platform has generated more than US$300 million in incremental customer revenue and achieved average conversion-rate improvement above 20%. It also describes large lifts for named clients, including Dropbox, Replit and Cash App.
More than US$300 million in incremental customer revenue and over 20% average conversion improvement. Both results are Coframe-reported estimates in its November 9, 2026 announcement; their attribution methodology and independent audit were not supplied.
For marketers, the missing details matter as much as the headline figures. A conversion lift from one landing page may be difficult to extrapolate across an entire website. Results can be influenced by traffic mix, seasonality, campaign spend, baseline performance and the duration of an experiment. Revenue attribution is especially sensitive to how the counterfactual is constructed.
Business Insider's November 9 reporting confirms the investment and Coframe's broader pitch, but the reported commercial performance remains company-provided. Growth teams should ask to see experiment design, holdout methodology and deployment history before translating those claims into expected returns on their own sites.
What marketing teams should test before handing over the website
The clearest near-term opportunity is operational. A large team can spend weeks coordinating routine web tests, particularly when copy, design, development, legal review and analytics belong to different owners. Automating parts of that process could increase the number of ideas examined without increasing the same amount of production work.
But website optimization is not only a question of maximizing clicks. An AI system can unintentionally favor misleading urgency, inaccessible layouts or aggressive personalization if those tactics improve a narrow metric. Teams need constraints for brand consistency, user consent, accessibility and legal claims before allowing software to alter live customer experiences.
Buyers should also separate the system's ability to generate variations from its ability to measure them correctly. Useful evaluation questions include how visitors are randomized, whether segments are compared fairly, which changes require approval, how a version can be rolled back and who owns the experiment data. Those controls determine whether automation adds reliable learning or simply produces more activity.
The new funding gives Coframe resources to pursue a more automated website stack. For marketers, the relevant test is narrower: can it shorten experiment cycles while preserving evidence quality and customer trust? Until the claimed gains are reproducible under clear measurement conditions, the strongest takeaway is a faster way to test, not a guaranteed faster way to grow.
