From a 25 to Agent-Native in one evening.

Initial condition

On the morning Google rolled out its May 2026 Core Update, we ran tricitiesailab.com through isitagentready.com, a free scanner Cloudflare launched to score how well AI agents can find and use a website, on a 0-to-100 scale. We expected a decent result. We got a 25.

Constraints

• The fix had to happen same-day, alongside a full client workload.
• No rebuild or redesign was on the table. This had to work with the existing site.
• Whatever we did had to hold up under Cloudflare's own scoring criteria, not just look right to a human.

What was changed

• Rewrote the bot access rules so search and AI-citation crawlers (ChatGPT, Claude, Perplexity, Google's AI systems) are explicitly allowed, while bots that scrape content only to train models are explicitly blocked.
• Added a machine-readable, markdown version of every page, served automatically when a requesting agent asks for it, so AI tools can read clean content without parsing navigation, images, or layout markup.
• Published a machine-readable "site card" and API catalog: a structured summary of what the business does, what services it offers, and how to find more, aimed at software agents rather than human visitors.

Tools used

Cloudflare's isitagentready.com scanner for scoring; standard web infrastructure (robots.txt, HTTP content negotiation, and the emerging llms.txt / MCP server-card conventions) for the fixes themselves. No new platform or vendor was introduced.

What was human-reviewed

Every rule change was reviewed manually before publishing, in particular the bot-access policy, since getting it wrong could either block legitimate AI citation traffic or hand training bots free access to the site's content. The scan results were re-run and read by a person after each round of changes rather than trusted blindly.

Result

By that evening, tricitiesailab.com ranked "Agent-Native," the top tier on Cloudflare's scale. As of July 20, 2026, the site scores 100 on the content-scope scan (discoverability, content format, and bot access checks); API and commerce categories are outside the scope of a content site and were not scanned. See the methodology page for the full scan history and exact dates.

Limitations

This scanner measures technical readiness: can an AI agent access, parse, and understand the site. It does not measure whether AI tools actually recommend the business in a real answer, which also depends on entity authority, reviews, and citations elsewhere on the web. A high technical score is necessary but not sufficient for AI visibility.