SMS verification for AI agents

You've built an agent that needs to sign up for a SaaS tool — say, to file a support ticket on behalf of a user, or to pull data from a service that gates its API behind a free-tier signup. The signup form asks for a phone number "to send a verification code." That's where most agent workflows stall. SIMRelay is the path through.

How it works in an agent loop

Your agent claims a hosted SIM number from the SIMRelay API (programmatically — no human in the dashboard). It uses that number to fill the signup form. The verification code arrives at the number, gets delivered to your endpoint via signed webhook (or polled via API), and your agent reads it, submits it, and continues the flow.

From the service's perspective, the phone number is a real, working SIM that passes their carrier-lookup checks. From your perspective, you've turned "SMS verification" from a wall into a one-line API call.

Programmatic claim and release

Numbers can be claimed for a specific agent run, locked for the duration so no other process intercepts the SMS, and released back to the pool when done. Or kept long-term if your agent needs a stable identity (see phone-number identity).

Receiving the SMS

Three ways to get the verification code into your agent:

  1. Signed webhook — fastest, push-based, requires an HTTPS endpoint.
  2. API polling — easiest to integrate, slightly higher latency, no server needed.
  3. MCP tool — for agents running on Claude / Cursor / any MCP-aware host. See the MCP server.

What you're not getting

SIMRelay does not bypass anti-abuse systems. Services that specifically detect agent-driven signups (CAPTCHAs, fingerprinting, behavioral signals) will still detect them — we just remove the phone-verification bottleneck. Plan your agent runs accordingly: real SIMs solve the phone-number problem, not the broader bot-detection problem.