What does a policy-as-code policy contain?

A policy is JSON describing the outcome categories, the criteria the model checks, and per-criterion severity and required flags. You save it, get a stable policy ID, and pass that ID to the verify endpoint with each image.

Why use policy-as-code instead of a fixed model?

Because rules change. Policy-as-code lets you edit, diff, and version the verification logic — tuning it per city or workflow — without retraining a model or updating hardware. The same API then enforces the new rules immediately.

Can I start from an existing policy?

Yes. The policy template gallery includes production-modeled policies for micromobility parking, designated-bay compliance, vehicle damage, and proof of delivery. Copy one, edit the wording, and ship it.

Glossary

Policy-as-Code

Policy-as-code is the practice of expressing a verification rule set — the categories, the per-criterion checks, the severity, and what gates a pass or fail — as plain, editable, version-controlled configuration rather than hard-coded or opaque model behavior.

In VerifyAI, a policy is a structured JSON object: a list of outcome categories (for example "Good Parking" / "Bad Parking" / "Poor Photo"), a list of criteria the model evaluates ("vehicle clearly visible", "not blocking sidewalk", "not in roadway"), and per-criterion settings for severity and whether the criterion is required. The same policy object is what you save and then reference by ID when you call the verify endpoint.

Treating the rules as code has practical benefits. You can diff a policy in code review, version it alongside the rest of your stack, and tune it city-by-city or SOP-by-SOP without retraining a model or shipping new firmware. When a city changes its designated-parking requirement, you edit the policy text — not the integration.

Policy-as-code is what lets one verification API serve very different jobs from the same plumbing: a parking policy, a vehicle-damage policy, and a proof-of-delivery policy are all just different configurations. Ready-made starting points live in the policy template gallery, and the city-parking-policy guide shows how to encode designated-zone and corral rules.

Learn more

Related terms

Photo Verification API

A photo verification API is a service that accepts an image plus a policy and returns a structured verdict — typically pass/fail with the reasons and a category — confirming that the photo shows a real-world condition or that it meets a defined rule.

End-of-Ride Verification

End-of-ride verification is the check a shared micromobility operator runs when a rider ends a trip: the rider photographs the parked vehicle, and the photo is verified against the city's parking rules before the ride can close.

See it on a real photo

Start free in the sandbox — $5 in credit, no card required. Verify your first image against a policy in minutes.

Start free in the sandbox Read the quickstart

Learn more

Related terms

See it on a real photo

Get in Touch