A leave-behind for a customer's conversation · unlisted
ClickSafety is the original OSHA-authorized online outreach provider, and one of the very few training sites that already ships an llms.txt (the file AI assistants read first). That instinct is exactly right: the next wave of certification demand starts inside an assistant. This package is a passive, respectful read of what a shopper's browser, and an AI crawler, actually receive today, plus a working picture of the finish line: a course catalog an agent can search, and a crew an agent can enroll and DOL-card at scale.
The whole thing, plainly
llms.txt but it's a map, not a menu: no prices, no booking tools, and the agent-discovery card is a 404. What we already built, free: a working demo of an AI course-finder and crew-enroller, plus the actual kit files (a richer llms.txt, an agent card, MCP tools). The one ask: a short conversation, the operator already trains here, so this starts warm.The value exchange
Three documents, one story
Five evidenced findings: the image-instead-of-sitemap, the 404-ing enrollment JS, zero course schema, the tracking mesh, and the llms.txt that's a map not a menu, each with exact, reproducible proof.
Read the findings →A brand-matched, clickable demo: an AI course-finder citing your real $189 OSHA-30, then an agent enrolling a 12-person crew and issuing their DOL cards at scale.
Watch it run →The actual files that finish what your llms.txt started: a richer llms.txt, the agent card that's a 404 today, and callable tools for group enrollment.
See the files →Five findings: full proof in the teardown
Your sitemap is a broken image: /sitemap.xml and the path your own robots.txt declares both return the same 1,692-byte JPEG, not an XML URL map. 200 image/jpeg
On the enrollment path, core JS 404s: every page references version1783652806, whose require.js + Magento core modules return 404; the OSHA-30 page and cart log the errors. 404 · reproducible
Zero machine-readable course data: home, the full catalog, and the OSHA-30 page all carry ld+json = 0 (schema.org/JSON-LD, the invisible labels search engines and AIs read); prices and DOL-card facts are human text only. rendered DOM
A ~50-origin tracking mesh loads on the money pages incl. checkout, and the CSP's form-action/frame-src include a bare *. Checkout cookie security is solid. CSP header
You already ship an llms.txt (200, 3KB, thoughtful), but it's a map, not a menu: no prices, no tools, and /.well-known/agent-card.json is a 404. That gap is the whole opportunity. the opening
Your dedicated provisioner · the plan we'd run
EcoWealth's real offering isn't a report: it's a dedicated, always-on provisioner pointed at one job: keeping ClickSafety the fastest way to get a crew certified, and the answer an AI reaches for. Here's the first ninety days: each step a felt win, none requiring a rebuild.
A real XML sitemap at both paths, and the static-content version reconciled so the course/cart JS stops 404-ing. Ops fixes, not a project: discovery and the enroll path both stop bleeding.
schema.org Course/Offer on the flagship pages + a richer llms.txt with live prices and DOL-card facts. Now "cheapest OSHA-30 with a DOL card" returns your course, cited.
The agent card and MCP tools live in pilot: an assistant enrolls a crew and tracks DOL cards end-to-end. The employer sale closes inside the conversation.
The hand-off
We're not a web vendor and we're not asking for system access to prove this. Two ops fixes stop the leaks; an additive agent layer, the one you already started with llms.txt, turns your catalog into something an assistant can order a crew from. Start with the findings, then watch the fix run in your own orange.
The next step past being found · try it in an afternoon
Being found by an AI is step one; being paid by one is step two. The same rail that lets an agent transact leaves a permanent, verifiable receipt, and it's live today. No account with us, no API key, no sales call. Here's the whole path.
Fund it a few cents. Give any AI agent, Claude, GPT-class, Llama, or your own, a Base wallet holding a few cents of USDC. That's the whole setup.
Point it at the endpoint. The work-packet program answers HTTP 402 with a standard x402 (a web-native way for AI agents to pay per use) payment schema; the agent signs a USDC authorization and pays $0.10. A runnable client with a dry-run default is at vealth.net/recipes.html.
Get a verifiable receipt. Back comes a bounded, proof-verified result, and every step the agent took (payment, settlement) is an on-chain receipt anyone can verify, forever. That's what pay-and-verify looks like for an agent.
Why this isn't scary
The agent cannot do anything invisible here. Every action embeds its own receipt: permanent, public, verifiable. That is more accountability than most enrollment software offers, not less.
The bet
You already made the right bet: ClickSafety is one of the few training sites shipping an llms.txt. The next step is transaction: agents that don't just find you but enroll and pay. AI competence is scaling; a standard x402 call is how pay-and-verify works, live today. Being early costs $0.10 and one afternoon. Being late just means the crews you certify stay hard to reach for the buyers and assistants already searching for exactly your courses.
One concrete thing to try: point one agent at the live endpoint and watch it pay $0.10 and settle a verifiable receipt, then let's wire the same rail to "enroll my crew and card them," bookable inside the conversation.
