Behavioral pool routing
The behavioral IP segregation model that routes sends between behavioral pools — Clean and Mixed today, an explicit Outbound pool in Phase 2 — based on the classifier's read of the send's behavioral signals, not account settings.
Behavioral pool routingis the IP segregation model that decides which infrastructure every send uses. The classifier reads your send’s behavioral signals and routes sends between behavioral pools — the explicit Outbound pool ships in Phase 2. You do not pick the pool; the platform does.
The pools
- Clean. Shared IPs reserved for transactional-pattern senders. Low volume, known recipients, stable engagement, no cold-script content. Inbox placement is excellent because every sender on Clean looks legitimate to receiving mail servers.
- Mixed.Shared IPs for senders with mixed or unclear behavioral signals — new agents building reputation, warm-prospecting with some cold-pattern signals, or agents that deviated from their baseline. Not a penalty tier; agents move to Clean as reputation builds.
- Outbound.Dedicated IPs for explicit cold B2B prospecting. Ships in Phase 2 (roadmap) with ~14-day IP warmup. Reputation is fully isolated — no shared-pool contamination in either direction.
Pools shape deliverability treatment— which IPs carry the send and how receiving servers see it — not price. Pricing stays flat metered ($0.001 per send, coming soon) regardless of pool placement.
How placement is decided
The behavioral classifierruns on every send before delivery. It scores five signal families and returns a pool placement, a 0–1 classifier_score, and the four top-weighted signals. A score near 0 routes to Clean; near 1 routes to Mixed (or, once it ships in Phase 2, the Outbound pool).
Cold-pattern senders that land on Mixed are auto-throttled to 50 sends per day until their behavioral reputation firms up. The throttle lifts automatically once engagement signals clear. Any sender crossing 0.3% complaint rate auto-pauses, regardless of pool.
Why pool segregation matters
On a single-pool platform, a spam campaign and a 2FA agent share the same IPs. Complaint accumulation from the spam campaign degrades inbox placement for the 2FA agent. Behavioral pool routing makes this contamination impossible — the spam campaign never touches Clean IPs, so transactional agents are isolated at the IP level, not just the account level.
Read the behavioral pool routing post for the full rationale, including how spam economics break structurally when the classifier makes abuse costly.
What to read next.
Built for agents.
Self-serve at every volume.
Public API opens Q3 2026. Drop ~6 lines into your agent and ship.
$ npm install @mailsai/sdk