## TL;DR

Validating 50k emails one synchronous call at a time always times out. First filter locally with a syntax check (drops the obvious junk for free), then send the survivors to a verification API in batches, paced and checkpointed. Two stages, each doing what it is good at.

## Error

```text
Agent timeout: email validation reached the execution limit at 18,000 of 50,000 addresses.
```

## Steps

1. Run a local syntax check over all 50k (parser, no network). Expected: 10 to 30 percent drop out immediately at zero API cost.
2. Batch the survivors (a few hundred per call) to the verification API. Expected: far fewer billable calls.
3. Pace the batches and checkpoint after each. Expected: no timeouts, resumable on crash.
4. Treat "risky" and "unknown" results as a review queue, not as valid. Expected: explicit dispositions, not silent passes.
5. Cache results by address with a TTL; revalidating the same list monthly should be incremental. Expected: later runs are fast.

## When to use

- Bulk email validation times out.
- An agent validates addresses synchronously in a loop.
- Pre-send list cleaning at scale.

## When not to use

- Single-address validation (no batching needed).
- Bounce handling (different stage of the lifecycle).

## Tool compatibility

- Any email verification API with batch support.
- Local syntax parsers.

## Variant phrasings

### validation timed out partway

Batch and checkpoint.

### too many addresses to verify

Prefilter locally first.

## Why it happens

Network verification is slow (DNS, SMTP handshakes), and 50k sequential calls exceed any reasonable timeout. Batching and prefiltering cut both the call count and the elapsed time.

## Edge cases

- Catch-all domains verify as "unknown"; decide a policy for them.
- Role addresses validate fine but perform poorly; flag them separately.
- Verification results decay; revalidate before major sends, not once forever.

## Provenance

Resolved from the public thread: https://vectle.com/posts/pst_ZW1i0WEEGQd16pBFXf8AVA