Cycling through all six. Tap any point to stop.

Measured on six things

Track and improve your brand's visibility across ChatGPT,...

Speed

Lighthouse mobile 58/100 with a 13.6s LCP; desktop is worse at 48/100. CLS is a clean 0 and FCP is 1.7s, but the largest element takes over ten seconds to paint on mobile.

Screenshots

1 / 1

Refine AI

About this app

Refine AI is an AI visibility tracking platform that monitors how your brand is mentioned and recommended by leading AI assistants like ChatGPT, Claude, Gemini, and Perplexity. It offers a suite of AI agents that analyze brand knowledge, research prompts, identify citation gaps, and even generate content designed to improve your AI search presence. With features like ChatGPT Ads tracking, competitive intelligence, and a custom MCP server for integration with AI assistants, Refine AI provides a comprehensive solution for marketing teams, SEO professionals, and brands looking to win in the new era of AI-driven search.

Key features

1AI visibility tracking across ChatGPT, Claude, Gemini, Perplexity and Google AI Overviews with daily reports and competitor intelligence.
2Agent suite: Brand Knowledge, Prompt Research, Citation Gap, Content, Reddit and Site Health agents working from your own data.
3Custom MCP server for Claude, Cursor and ChatGPT — query visibility, manage prompts and publish content from the chat.
4ChatGPT Ads tracking (early access) showing which brands buy the sponsored slot on your prompts.
5Content publishing to WordPress and Shopify blogs, plus AI Feeds answer pages for prompts you lose.

Target customer profile

Drawn from the product itself, not from a survey.

Demographic

Marketing teams and brand managers

Pain points

Uncertainty about brand visibility in AI answers; competitors appearing in AI recommendations

Primary needs

Monitor brand mentions across AI platforms, track sentiment, and get actionable insights to improve presence

Demographic

SEO professionals and agencies

Pain points

Traditional SEO tools don't cover AI search visibility

Primary needs

Integrate AI visibility into their service offerings, track prompts, and measure ROI

Demographic

E-commerce brands and founders

Pain points

Products not being recommended by AI assistants; lack of time to manage AI search

Primary needs

End-to-end AI visibility management, including product page optimization and content creation

Scout Review

Written by AI from measured evidence, scored out of 100.

Refine AI is a competent, clearly positioned AI visibility tracker with a genuinely interesting MCP server and a real agent suite behind it. The landing page is polished, the onboarding is conversational, and the pricing is transparent (€165/€321/custom). The problems are measurable: a 13.6s mobile LCP, a 4/12 security header checklist with HTTPS not enforced, and three accessibility audits failing. It's a solid mid-pack product in a crowded category, with one hook (MCP) that could actually differentiate it if the team leans into it.

Measure by measure

Usability

7-day trial with no card, results in 24 hours, and a conversational onboarding that fills your brand profile from your site. Undercut by a demo CTA fighting a smaller 'Start my audit' in the nav.

Design

Coherent orange accent, clean agent-card grid, and a legible dashboard with a ranked competitor table. Held back by a hero mockup that reads as empty white space and dense small text in agent cards.

Speed

Lighthouse mobile 58/100 with a 13.6s LCP; desktop is worse at 48/100. CLS is a clean 0 and FCP is 1.7s, but the largest element takes over ten seconds to paint on mobile.

Security

HTTPS not enforced, no CSP, no frame protection, no X-Content-Type-Options, no Referrer-Policy. HSTS, SPF and DMARC are present, but the 4/12 header checklist and absence of any SOC 2 or pen-test claim keep this mid-band.

Accessibility

Lighthouse accessibility 92/100 with a responsive layout and clear heading hierarchy. Three failing audits (contrast, touch targets, accessible names) and no WCAG statement or multi-language support keep it out of the top band.

Growth

Real, growing problem (brand visibility in AI answers) with a specific wedge: an agent suite plus ChatGPT Ads tracking and an MCP server. Differentiated, but the category is crowded and the moat is thin.

Improvement areas

Mobile LCP of 13.6s and desktop Lighthouse 48/100 — the hero render is the bottleneck.
Security headers at 4/12: HTTPS not enforced, no CSP, no frame protection, no Referrer-Policy.
No SOC 2, ISO or pen-test claim despite requesting Search Console and GA4 access.
Three failing accessibility audits (contrast, touch targets, accessible names) and no WCAG statement.

Design & Usability

Refine AI's design is competent and coherent: an orange accent used sparingly, a clean 2x3 agent grid, and a dashboard screenshot that actually shows a legible line chart and ranked competitor table. The weak spots are a hero mockup that reads as an empty white card and agent cards with dense small text. Usability is stronger than the design: a 7-day trial with no card, first results in 24 hours, and a conversational onboarding that reads your site and fills the brand profile. The main friction is CTA competition — 'Book a demo with the founders' and 'Start my audit' both fight for the same click in the nav and hero.

Speed & Security

Speed is the weakest measured dimension. Lighthouse mobile scores 58/100 with a 13.6s LCP, and desktop is worse at 48/100 — the largest element takes over ten seconds to paint on mobile, which is well past the 2.5s 'good' threshold. CLS is a clean 0 and FCP is 1.7s, so the bones are fine; the hero render is the problem. Security is mid-band: HSTS, SPF and DMARC are present, but HTTPS is not enforced, and there's no CSP, frame protection, X-Content-Type-Options or Referrer-Policy. The header checklist scores 4/12, and no SOC 2, ISO or pen-test claim appears on the site — a gap for a product that asks for Search Console and GA4 access.

Accessibility & Growth

Accessibility scores 92/100 on Lighthouse with a responsive layout and clear heading hierarchy, but three audits fail: contrast, touch targets, and visible labels not matching accessible names. There's no WCAG statement or multi-language support. Growth is a judgement on the idea: AI visibility tracking is a real, growing problem, and Refine's wedge — an agent suite (Brand Knowledge, Prompt Research, Citation Gap, Content, Reddit, Site Health) plus ChatGPT Ads tracking and a custom MCP server — is more specific than most. But the category is crowded with Profound and Athena pitching the same dashboard, and the MCP hook is the only structurally hard-to-copy piece. The domain is ~0.7 years old with only LinkedIn as a social signal, so traction is early — context, not the score.

Conclusion

If you're a marketing team already paying for SEO tooling and want to see where ChatGPT and Perplexity actually rank you, Refine is a reasonable trial — the 7-day no-card window and 24-hour first results make it cheap to test. If you're security-sensitive or need enterprise compliance, wait: no SOC 2, no enforced HTTPS, and a thin header posture. The MCP server is the reason to watch this one; the rest is table stakes in a category that's filling up fast.

Suggestions for improvement

Fix the 13.6s mobile LCP — compress or lazy-load the hero mockup; desktop scoring 48/100 is a red flag.
Enforce HTTPS and add CSP, frame protection, X-Content-Type-Options and Referrer-Policy to move past 4/12.
Resolve the three failing Lighthouse a11y audits: contrast, touch targets, and accessible-name mismatches.
Pick one primary CTA — 'Book a demo' and 'Start my audit' competing in the same viewport splits intent.

How we score

Alternatives comparison

Named competitors, point by point. Nobody paid to appear here or to be left out.

	Refine AI	Profound	Athena
AI engines monitored	ChatGPT, Gemini, Perplexity, Claude, Google AI Overviews (Autopilot+)	ChatGPT, Perplexity, Gemini, Claude, Copilot (per public site)	AI search engines (specific list not verified)
Prompt tracking limits	50 (Visibility), 200 (Autopilot), unlimited (Scale)	Tiered by plan (specific limits not verified)	Tiered by plan (specific limits not verified)
Content generation / publishing	Content agent, Reddit agent, publish to WordPress and Shopify	Content optimization features (publishing integrations not verified)	Optimization features (publishing integrations not verified)
MCP server / assistant integration	Yes — custom MCP server for Claude, Cursor, ChatGPT	Not advertised on public site	Not advertised on public site
Free trial	7-day free trial, no card required	Demo-led (free trial not verified)	Demo-led (free trial not verified)
Entry pricing	€165/mo Visibility; €321/mo Autopilot; custom Scale	Custom / demo-led (not published)	Custom / demo-led (not published)