Coframe is a startup that offers conversion rate optimization for their customers. Their services are similar to what a traditional CRO consultancy offers, except their optimizations are generated by LLM models like ChatGPT.
What is AI CRO?
This is one example of a larger trend called AI CRO. Normally, working with a conversion rate optimization consultancy is high-touch and involves lots of back and forth. Consultants might cost upwards of $30,000 per engagement, while they still can't directly land changes to your website because they might be blocked by engineering resources or access limitations (due to regulatory processes, for example). AI CRO sidesteps these problems by delivering the same value of CRO (A/B tests and improved conversion tactics) at a lower cost and lower friction.
Coframe's landing page says:
Launch tests in hours, see conversion lift in 2 weeks. No lift required from your team.
Coframe is one of the first startups to have offered AI CRO as a service. Coframe is a contact sales only process, meaning you can't self onboard your app or website onto their product.
Alternatives to Coframe
1. Atonomo for AI CRO
Atonomo is an AI product manager that identifies and applies AI-generated conversion rate optimizations autonomously. Atonomo plugs into your existing product analytics and experimentation platform to learn about your users and suggest ways to better convert and retain them.
While Atonomo and Coframe are similar, there are some important differences.
-
Atonomo ingests more complex data from PostHog, MixPanel, and Amplitude, and can watch session replays of users for inspiration. Coframe mostly tracks button clicks and user properties like browser.
-
Atonomo uses your existing experimentation platform as the record for A/B tests. If you already use Growthbook or PostHog to run experiments, Atonomo works alongside you in the same workspace. Coframe introduces a new record for experiments.
-
Only Atonomo can interface directly with a more complex application on GitHub by submitting pull requests against it, or can collaborate with a software engineer by giving a Claude Code or Cursor prompt to copy and paste to implement suggestions. Both Coframe and Atonomo have a JS snippet that lets you skip the engineering team at a cost of 200ms added latency on target web pages.
2. Running your own A/B test experiments
The most common alternative to using an AI CRO product like Coframe or Atonomo is managing the entire conversion optimization process in-house. This requires generating ideas, iterating on them, and implementing the experiments, which is normally a cross-team collaboration between marketing, growth, and engineering teams.
When teams run their own A/B tests, they normally run them on A/B test platforms like Optimizely, StatSig, or Growthbook. This is in addition to data pulled from product analytics like MixPanel and PostHog.
Most teams start out by running their own A/B tests, and only decide to adopt AI CRO tools when they want to take growth seriously. Any AI CRO product will greatly increase the rate at which you start and run A/B tests.
What is Coframe's relationship with OpenAI?
Coframe fine-tuned a model with OpenAI one year ago. This type of co-promotion work with an AI lab like OpenAI often involves paying OpenAI. The Coframe website doesn't actually say that OpenAI is a customer and the Coframe JS snippet is not live on the OpenAI landing pages we checked.
Because of how rapidly LLMs have improved, it's likely this model they finetuned is much worse than state-of-the-art models now, like Claude Opus 4.5 or Google's Gemini 3 Pro. It's not clear whether this partnership continues, as of when we looked in January 2026.