Choose which AIs see your content. Prove it.

Your robots.txt is theater. The actual levers are somewhere else.

Most sites believe they've managed AI access because they edited robots.txt. Robots.txt is a polite request. Compliant crawlers honor it, sort of, and non-compliant ones don't read it. Neither case produces evidence.

The load-bearing mechanism is the verified-bot firewall rule inside your CDN. It checks User-Agent and reverse-DNS against the vendor's published IP ranges. Cloudflare exposes it as cf.client.bot. Every serious CDN has an equivalent. Almost nobody is using it on purpose.

We proved the mechanism on ourselves. Here's what 14 days of one misconfigured firewall rule looked like:

One vendor starved. Three untouched. One missing firewall clause. We didn't set out to demonstrate the mechanism — our own auto-malicious-IP rule drifted over Bingbot's range and ChatGPT stopped seeing us. That diagnostic trail is the product.

Read the full write-up with receipts →

Four tiers. Pick the work you actually need.

Each tier is the previous tier plus the next layer of work. No feature-fencing games. If you need it, it's in the tier named after it.

Who built this, and why you should believe the receipts.

We are DugganUSA LLC — a two-person shop in Minneapolis that built a commercial HAIC benchmark for AI brand perception, audits 250+ domains on a public leaderboard, and publishes a free STIX threat feed. No patents are filed and no SOC 2 report exists; the checkable work is public.

When our own site lost 86% of ChatGPT traffic in two weeks, we had the telemetry to see it, the audit log to prove it, and the diagnostic muscle memory to fix it the same night. That's what we sell.