Determine the required assurance level: age estimation from selfie, age inference from credit bureau data, or hard age verification via document date of birth
For document-based age verification, submit the document to an IDV vendor and extract the date of birth field from the verified document data
Calculate the user's age at the time of the transaction by subtracting the extracted date of birth from the current date
For AI-based age estimation, submit the selfie to the age estimation endpoint and receive an estimated age range with a confidence score
Apply your minimum age threshold against the verified age or the lower bound of the estimated age range depending on the assurance model chosen
Store the age verification outcome, method used, and timestamp for compliance reporting without retaining unnecessary biometric data beyond the required retention period
Known gotchas
AI age estimation has non-trivial error rates especially near boundary ages (e.g., 17 versus 18); regulators in some jurisdictions do not accept estimation alone for hard age-gated content
Date of birth extracted from an IDV check is only as reliable as the document itself; expired or potentially fraudulent documents reduce confidence
Different jurisdictions have different age thresholds and different acceptable verification methods; do not apply a single global policy without jurisdiction-specific review
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp