Infrastructure
When a provider pushes back, we slow down. We do not retry harder.
We operate our own mail infrastructure, so the send path can react to what receivers are telling us. The deliverability work most teams do by hand (watching dashboards, pausing a domain, shifting volume) happens automatically, per destination.
- Per-destination circuit breaking
- Throttling that responds to signals
- Automatic routing to healthier IPs
- Pool or Dedicated, same API
How it works
The send path has a feedback loop
Most providers accept your message and retry on a fixed schedule. Ours watches how each destination responds and changes its own behavior. That is the part a reseller cannot build.
It notices per destination
Delivery outcomes are tracked separately for every recipient domain and every sending IP. Gmail being healthy tells us nothing about Outlook.com, so we do not average them together.
It slows that path down
Sustained deferrals from one destination tighten the pace to that destination only. Your mail to everyone else keeps moving at full speed.
It stops retry storms
A path that keeps refusing connections is paused instead of hammered, then probed occasionally until it recovers. Pausing does not burn a message’s retry budget.
It routes around trouble
When one IP is struggling with a destination, queued mail moves to an IP with a better record for that destination. Chosen from evidence, not round-robin.
In practice
Built for the moments that damage reputation
Rate limits, retry storms, and one-provider problems are where good senders get hurt. The send path is designed to respond the way a careful operator would, without you having to be online when it happens.
The risk
Persistent 4xx deferrals during a ramp are a rate limit, not a bug. Pushing harder makes it worse.
IP warmup guide →What we do
We tighten the pace to that destination automatically, then earn the speed back as clean delivery resumes.
The risk
Retrying into a throttling receiver makes the throttling worse, and burns the reputation you are trying to build.
Email infrastructure guide →What we do
Repeated connection-level failures pause that path and probe it on a schedule, so retries cannot escalate the problem.
The risk
A problem isolated to one provider has a different cause than a problem everywhere. Blending them hides the signal.
Deliverability guide →What we do
The send path is keyed by recipient domain, so a single struggling provider is contained instead of shaping your whole reputation.
Lanes
Shared transactional pool, or your own servers
The same reactive send path runs on both. Dedicated gets its own servers, IPs, and independent reputation: built, warmed, and managed by us.
Pool
Shared, transactional only
- Paying customers only — no free-tier noise
- No marketing blasts or cold outbound
- Warmed, monitored, and operated by us
Dedicated
Your own servers and IPs
- Provisioned for your workspace alone
- IPs listed in the dashboard for allowlists
- Built, warmed, and managed by us
Same API either way — moving to Dedicated does not change your send code.
For the curious
What this means in practice
A little more detail if you are evaluating whether we run mail infrastructure or resell someone else’s.
- Health is measured per sending IP and per recipient domain, over a recent window, not as one global success rate
- New destinations start at a cautious pace, and gateways get more caution than large consumer providers
- Speed is earned back through sustained clean delivery, and a fresh deferral steps it back rather than resetting it
- A pause backs off further each time a path fails its recovery probe, so a genuinely dead route stops consuming attempts
- Bad recipient addresses are separated from destination pushback, so one dead mailbox never slows a domain
- Pacing pushes queued mail forward quietly. It is our pacing decision, not a recipient-side delay we report as one
None of this is configuration you have to manage. We tune it on the sending lanes we operate, and the deliverability endpoint reports what the path is seeing on your behalf.
Send on infrastructure that reacts on its own
Start free on Sandbox, move to the transactional Pool for production, and upgrade to Dedicated servers and IPs when you need them. Same API throughout.