No ticket queues, no chatbots pretending to be humans. Your message lands with a named member of our engagements team — and if you're carrying a model-safety problem, it gets read today.
All channels open · Avg first response 3h 12mFrontier labs, medical, legal, finance AI, and SaaS teams scoping evaluation programs, preference data, or red-teaming. Bring your failure surface; we'll bring the cohort plan.
enterprise@judgemyai.comPhDs, MDs, JDs, and engineers applying for highly-paid remote AI evaluator and trainer roles. Credential questions, calibration process, cohort availability.
talent@judgemyai.comJailbreaks, prompt-injection families, data leakage, and alignment failures found in the wild. Coordinated disclosure, written remediation status, PGP available.
security@judgemyai.comJournalists citing our research, teams requesting aggregate findings, and conferences seeking red-team methodology talks. Attribution guidelines on request.
press@judgemyai.comPrefer a form to email? Our encrypted intake module loads in this panel — the fastest path from first message to scoping call. Until it appears, the direct lines above are quicker anyway.
A real human confirms receipt, identifies your domain, and assigns a named engagement owner. No auto-responder theater.
Within 4 business hoursA 30-minute confidential call to map your model's failure surface — under mutual NDA if you prefer, before anything sensitive is shared.
Scheduled at your paceWe return a short written assessment: where your exposure is, what evaluation approach fits, and what it would take to calibrate a cohort.
Written, reviewable, yoursA scoped proposal with expert composition, timeline, deliverables, and pricing — the same document our frontier-lab clients sign.
Launch-readyEnterprise and lab engagement inquiries are answered within 4 business hours, and every inbound message is assigned a named human owner within 24 hours. Security disclosures are monitored around the clock.
Yes. A mutual NDA can be executed at first contact — ours or yours — before any model, benchmark, or failure data is shared.
Yes. Scoping scales to budget, and a subsidized open-license program exists for community model developers, including open-released preference and safety data sets.
Email security@judgemyai.com, monitored 24/7 under coordinated disclosure. PGP encryption is available on request, and verified reports receive a written response with remediation status.
JudgeMyAI operates as a fully distributed network of vetted experts across 40+ countries. Engagements run asynchronously with scheduled calibration calls aligned to your time zone.
Whether you're shipping a model that cannot fail or you're one of the rare minds qualified to judge one — we've been waiting for your message.
AI doesn't improve itself. Humans do. The ghost in the machine.