Credentialing 1,870 nurses, physicians, and medical coders takes more than a checkbox. Here's the full verification flow for the Medical guild — and what it means for teams building medical AI.
Why medical annotation is different
Medical AI development requires annotators who understand the clinical context of what they're labeling. An image of a pulmonary nodule labeled by a general crowdworker and the same image labeled by a board-certified radiologist are not the same thing — even if the label value is identical. The confidence, the edge-case judgment, the ability to flag ambiguity rather than guess: that's what clinical expertise produces.
FDA guidance on AI/ML-based software as a medical device increasingly asks developers to document the qualifications of the humans who created their training data. The Medical guild gives you that documentation.
The verification process
- License number submission — cross-referenced against state medical board registries in real time
- Active status check — licenses must be current, not expired or suspended
- Specialty classification — we map each worker to the task types appropriate for their credential
- Domain assessment — a short set of cases reviewed by a Guild Master before approval
- 48-hour SLA — applications are processed within two business days
What Medical guild workers can do
Approved Medical guild workers are matched to tasks based on their specialty. A radiologist sees imaging annotation tasks. A clinical coder sees ICD-10 classification tasks. A pharmacist sees drug interaction review tasks. The matching is automatic and based on the specialty data collected at verification.
Always human-required
All judgment tasks routed to the Medical guild are system-enforced as human_required: true. The AI fallback will never fire for medical tasks. If no human completes within 24 hours, the task fails and the developer is not charged. There is no way to override this — it is a non-negotiable platform policy.
Getting access
The Medical guild is currently in private access. Teams building clinical AI products can apply via the Enterprise contact form. General availability is planned for Q2 2026.