All systems operational · 99.99% uptime

Webhooks that don’t page you at 3am.

Relay is the delivery layer for outbound events — fan-out, retries, signing, and replays behind one durable API. You send the event once; we make sure it lands.

Start free →Read the docs

$ npm i @relay/node · trusted by 4,200+ engineering teams

deliver.sh

# send an event to every subscribed endpoint
$ curl -X POST https://api.relay.dev/v1/events \
    -H "Authorization: Bearer sk_live_9f2a" \
    -d '{"type":"order.paid","id":"ord_9f2"}'

# 202 Accepted
{
  "id": "evt_2ZxQ7hK",
  "status": "queued",
  "endpoints": 3,
  "signature": "hmac-sha256",
  "attempt": 1
}

One POST in.
Guaranteed delivery out.

You stop writing retry loops, dead-letter queues, and signature code. Relay owns the hard, boring middle so your product code stays about your product.

Send us the event

A single authenticated POST. No queue to provision, no consumer to run. We accept it in under 20ms and take ownership.

// once, from anywhere relay.events.create({ type: "order.paid", data: order })

We fan out and retry

Every subscribed endpoint gets its own attempt with exponential backoff for up to five days. One slow customer never blocks the others.

// per-endpoint, automatic attempt 1 → 500 retry in 5s attempt 2 → 500 retry in 30s attempt 3 → 200 OK delivered

You verify one clean webhook

Each request is signed with a per-endpoint HMAC secret. Verify in one line, trust the payload, move on.

# your handler relay.verify(body, sig, secret) # throws on tampering, else returns event

Replay anything, anytime

Every event is retained and re-sendable. Ship a new endpoint on Friday, replay last week’s traffic into it, and it is caught up.

# backfill a new subscriber $ relay replay --since 7d \ --endpoint ep_new

Observability

Every attempt, logged and searchable.

A live delivery monitor for each endpoint — success rate, latency, and the full request/response of every attempt. When something fails, you already know why.

99.99%

Delivered, 30-day

42ms

p50 accept latency

Retry window

Deliveries · last 12hLIVE

order.paid1,204 delivered

invoice.finalized892 delivered

subscription.updated3 retrying

refund.created211 delivered

Specification

regionsus-east-1, eu-west-1, ap-southeast-1 — active-active, no primary

throughput50,000 events / second sustained per key, burst to 120k

retry_policyExponential backoff, jittered, up to 5 days or 24 attempts

signingHMAC-SHA256 per endpoint, rotating secrets, timestamp tolerance

orderingBest-effort by default, opt-in per-key FIFO with strict sequencing

sdksNode, Python, Go, Ruby, Rust, PHP — plus a raw HTTP + OpenAPI spec

retention90 days of full payloads, replayable; export to your own bucket

sla99.99% monthly delivery, with public status and incident credits

Priced per event, not per seat.

Start free. Scale to millions without a sales call. No overage surprises — hard caps if you want them.

Hobby

$0 / mo

For side projects and first integrations.

10,000 events / month
3 endpoints
24h retry window
Community support

Start free

Team

$49 / mo

Then $0.40 per 1,000 events over 1M.

1,000,000 events included
Unlimited endpoints
5-day retries & replays
FIFO ordering
Email & Slack support

Start 14-day trial

Scale

$299 / mo

Volume pricing from $0.18 per 1,000.

10,000,000 events included
99.99% delivery SLA
SSO & audit log
Bring-your-own bucket
Dedicated engineer

Talk to us