GROWAGENT

GROWAGENT is an AI conversion-rate-optimization (CRO) audit agent. You give it a domain, it audits the site the way a CRO specialist would, and it hands back a prioritized list of fixes with the code to make them.

The problem

CRO audits are slow and inconsistent. Two reviewers looking at the same page often flag different things, and a thorough manual audit of a whole site takes hours. The findings then land as a slide deck that nobody can act on directly.

What it does

• Finds the pages on its own. Give it a domain and it discovers up to 25 pages from the sitemap, robots.txt, and a homepage crawl, then lets you deselect any you do not want audited.
• Scores each page. Every page is graded against a 50-point CRO checklist across 10 categories, run in parallel batches so a full-site audit finishes in minutes.
• Looks at the page, not just the code. It reads the rendered screenshot with a multimodal model (Gemini 2.5 Flash) and applies named CXL persuasion and friction principles, so a recommendation can tell you which principle it came from.
• Checks the mechanics. Link health, calls to action, and form friction each get their own pass: broken links, dead CTAs, and high-friction forms are surfaced with the exact element and a fix.
• Hands work to a developer. Each recommendation ships with a code patch (HTML and Tailwind) and A/B test variations, and the whole report exports to nine formats.

The solution

The result turns a URL into an action list a developer can pick up the same day, with the reasoning attached to each item. It also learns: patterns it sees across audits feed back into later ones, so the recommendations get sharper the more it runs.