For the SheerID engineering & AI team · AI Gateway concept

Govern the AI behind every verification.

SheerID makes instant eligibility decisions at massive scale — and increasingly those decisions lean on AI for document checks, fraud signals, and edge cases. As that AI footprint grows across models and providers, it gets hard to govern: cost per customer, prompt and output logging, PII control on identity data, rate limits, and fallback. Cloudflare AI Gateway puts one control plane in front of all of it — on the network SheerID already runs its DNS on.

One pane across any model any provider every verification call

Verification is a trust product — and the AI inside it has to be just as trustworthy. When models touch identity data to make eligibility and fraud decisions, the questions get sharp fast: what is each model costing per customer and per check? Is PII being sent to a provider that shouldn't see it? Which calls could be cached instead of re-billed? Can you prove, for any decision, exactly what was sent to which model and what came back? Answering that across providers — without a governance layer — means custom plumbing rebuilt for every model and every integration.

Six controls over every model, from one pane.

Point existing model calls through AI Gateway — no model change, no re-architecture. You get:

Cost

Per-customer cost + spend limits

See exactly what each model, team, and customer is spending on AI — and cap it. Turn variable verification-AI cost into a predictable, attributable line item.

Caching

Response caching

Repeated and near-identical checks — common in high-volume verification — served from cache instead of re-billed to the provider. Lower cost and lower latency on the same decision.

🛡
Data protection

PII redaction + guardrails

Strip sensitive identity data before it reaches a model provider, and enforce guardrails on inputs and outputs — essential when the data being checked is personal by definition.

Observability

Full prompt/response logs

Every request and response logged across providers — the audit trail behind any AI-assisted decision, ready when a partner or auditor asks how a verification was made.

Resilience

Fallback + multi-provider routing

If a provider errors or rate-limits, route to a backup automatically. One integration point in front of every model, so adding or swapping providers is config, not a rebuild.

Control

Rate limiting

Per-key, per-model rate limits protect against runaway loops and abuse — the cost and stability guardrail under high-volume verification traffic.

You already run on Cloudflare.

SheerID's DNS already runs on Cloudflare — so AI Gateway isn't a new-vendor conversation. It's the same platform, one layer deeper: a control plane for the AI behind verification, alongside everything you already operate. Cloudflare also integrates with the identity and tooling stack your team already uses.

Protect margin

Per-customer cost visibility + caching turn variable AI spend into a managed, attributable line item.

Protect trust

PII redaction and full audit logs keep identity data controlled and every AI decision provable.

Move faster

One control plane instead of bespoke plumbing per model — more time on verification, less on AI infra.

Verification runs on AI. Govern it on one control plane.

SheerID is putting AI at the center of instant verification. AI Gateway is the layer that makes that AI governable, auditable, and cost-controlled at scale — with a one-line change to how each model is called, on the network SheerID already trusts. Let's keep the conversation going.

Matt Holscher · Cloudflare Digital Native team