ABM is one of the largest facility-services labor employers in the US: ~113,000 people, $8.75B revenue (FY2025, NYSE: ABM) cleaning, maintaining, landscaping, and electrifying buildings. It's genuinely ahead on machine-legibility (a live llms.txt and enforced email auth). This concept adds the layer above it: tamper-evident, machine-verifiable proof that the service was performed: the same primitive that turns grounds crews into restore-ladder work.
What we found: ABM already leads its sector on AI discovery, but four gaps sit above that baseline: a flat discovery file, self-attested proof of service, a grounds crew one step from ecological work, and sustainability pages with no proof-of-outcome layer.
What we built, free: a structured discovery file, a verified-receipt demo, and an agent-ready kit (agent card, callable tools, concierge script): all drafted from ABM's own public facts.
The smallest ask: one 10-cent automated payment call to try the live proof rail. No account, no sales call, nothing else requested.
/llms.txt (598 lines, 146KB, 200) and enforced email auth (DMARC p=reject, SPF -all). Genuinely ahead of the sector: credit first.grep -cE '^##' = 0. And service-page JSON-LD (the invisible structured labels search engines and AIs read) carries no Service/Offer schema.Baseline credited first: ABM is ahead. Then four evidenced findings, each with the exact URL, HTTP status, and re-check command: the flat-llms.txt / no-Service-schema gap, self-attested proof-of-service, the grounds-to-restore fit, and the PDF-bound sustainability commitments.
Read the brief →In ABM's own blue: a grounds crew completes a scope, captures photo+GPS+signature, and it settles to a tamper-evident receipt an enterprise client (or their AI) verifies independently: turning a self-attested timesheet into a sales asset at contract renewal.
Run the demo →Concept files: a structured llms.txt with real service sections, an agent card (the file that lets an AI assistant use you correctly), MCP (the protocol that lets an AI assistant call your tools) tool schemas for capability + service-proof verification, and a concierge persona: the natural next step on top of the discovery file ABM already ships.
0x76c17C…A14B (source verified on BaseScan), first fully settled packet workId 14 (2026-07-07); agent-payable tools over standard x402 (a web-native way for AI agents to pay per use: work-packet $0.10 · proof-check); live board counter at vealth.net/labor/stats (the counter is the truth, not this page).
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 enterprise software offers, not less.
The bet: ABM is already building for the agent era: it ships a machine-readable llms.txt, runs an "Inspiring Innovation" analytics program, and even keeps a live /ai-agent-training page (robots-disallowed, but there). This is the smallest possible extension of that. Being early costs $0.10 and one afternoon. Being late means the proof-verified track records belong to someone else.
The teach-block above is the door; this is the first real outcome. One bounded step, a rounding error against $8.75B, that returns a verifiable receipt tied to ABM's own "Advancing Sustainability" commitment:
work-packet endpoint. Keyless. You get a real packet object: the structured capability layer above your llms.txt, working in one call.Reason to use it: your service quality is currently a timesheet a client must trust. Each option ends in a receipt they verify themselves: a differentiator at renewal, and outcomes you can cite under "Advancing Sustainability."
