How YepIts.ai went from zero AI visibility to getting recommended

Using AIOptim's GEO methodology · 90-day dogfooding experiment

The problem

YepIts.ai is a YouTube video summarizer — paste any URL, get key takeaways and timestamps in seconds. It's a clean product with real users. But when we asked ChatGPT, Perplexity, Claude, and Gemini for "best YouTube summarizer," YepIts.ai was never mentioned.

Zero citations across all four models. The tools AI did recommend (Eightify, Notta, NoteGPT) weren't better — they just had stronger "entity presence" in the sources AI models crawl.

What we did (Week 1-2)

1. Technical foundation

• robots.txt — Explicitly allowed all 10 known AI crawlers (GPTBot, Claude-Web, PerplexityBot, Google-Extended, etc.)
• llms.txt — Deployed structured description following the llmstxt.org spec, giving AI models a concise, machine-readable summary of what YepIts.ai does
• Schema markup — Three JSON-LD blocks: Organization, SoftwareApplication (with offers and features), and FAQPage (7 Q&As)

2. Entity positioning

• sameAs links — Connected the domain to GitHub, LinkedIn, and other profiles so AI models can disambiguate "YepIts.ai" as a specific product, not a generic phrase
• Category clarity — Ensured every page and meta tag clearly positions YepIts.ai as "AI YouTube video summarizer" — the exact category query users type into AI models

3. Content & distribution (Week 2-3)

• Comparison articles — 3 honest comparison pages (vs Eightify, vs Notta, vs NoteGPT) with specific feature-by-feature tables — the format AI loves to extract
• "Best of" roundup — "Best YouTube Summarizer 2026" article positioning YepIts.ai alongside competitors
• Blog system — 6 SEO/GEO-optimized articles targeting AI-crawlable long-tail queries

Timeline & results

• Week 0 — Baseline
  Tested 20 queries across ChatGPT, Perplexity, Claude, Gemini. YepIts.ai: 0 citations. Competitors averaged 3-5 citations each.
• Week 1 — Technical build
  Deployed robots.txt, llms.txt, schema markup, entity profiles. No change in citations yet — AI models need time to crawl.
• Week 2 — Content deployed
  Published comparison articles, blog posts, "best of" roundup. Submitted to Google for re-crawling.
• Week 3-4 — Perplexity movement
  Expected: Perplexity (real-time web browsing) should pick up new content within 2-6 weeks. Monitoring daily.
• Week 8-12 — ChatGPT & Claude
  Expected: ChatGPT and Claude update their indices every 2-4 months. Citations expected to begin appearing as training data refreshes incorporate the new content.

What we learned

1. AI optimization is a different game from SEO. Ranking #1 on Google didn't help. The signals AI models use are completely different — entity presence, community mentions, structured data, and comparison content.
2. llms.txt is still early but worth doing. Only 7% of sites have adopted it. Low effort, potential first-mover advantage.
3. Comparison tables are critical. AI models extract structured data aggressively. Being in comparison tables on any high-DA site dramatically increases citation probability.
4. Timing matters. Perplexity responds in weeks. ChatGPT takes months. Setting expectations correctly is crucial for client satisfaction.

Want this for your tool?

We're offering the same methodology to SaaS founders. Full AI visibility audit, technical implementation, content creation, and monthly tracking.

Starting at $999 setup + $499/month — 3-5x below agency pricing.