Every channel. One signal.
Nervly is one API for SMS, WhatsApp, email, and push — with intelligent routing, provider failover, and built-in compliance across the regions you send to. You send. We deliver.
A message that doesn't arrive is a system that doesn't exist. Delivery is not a feature. It's the contract.
Cross-border messaging is held together with duct tape
6 APIS · 6 PAGERSAPI Fragmentation
Termii for Lagos. Twilio for Texas. SendGrid, FCM, APNs. Every provider is another API to learn, another outage to page you at 3 a.m., another deprecation notice in your inbox.
Hyperscale Bottlenecks
Flash sales, breaking news, mass OTP storms. Standard request/response architectures buckle exactly when your traffic — and your revenue — peaks.
The Regulatory Labyrinth
One misrouted SMS to a do-not-disturb number is a fine. One EU record on a US server is a violation. Compliance isn't documentation — it has to run in the hot path.
Twilio SendGrid WhatsApp Termii Infobip Africa's Talking Sinch Slack Firebase Apple Push Twilio SendGrid WhatsApp Termii Infobip Africa's Talking Sinch Slack Firebase Apple PushThat's why we built Nervly — the signal layer for the internet
Nervly ingests at hyperscale, routes on live cost and deliverability data, fails over across providers automatically, and clears every message against local regulation — before it leaves the building. One API. Every channel. Delivered.
Exactly-once, or not at all
Idempotency keys with configurable TTL mean a network retry never double-charges a customer or double-sends an OTP. Send with confidence at any scale.
Routing that does the math
Live cost arbitrage per destination prefix. WhatsApp first; if unread in 60 seconds, SMS fallback — automatically. You set intent. Nervly picks the wire.
Compliance in the hot path
Dynamic DND registry checks before dispatch. PII replaced with format-preserving tokens. EU data pinned to EU soil. Auditable by design.
What teams say after the first bad morning
Our provider had a bad morning and our OTPs just kept going. Nobody on my team found out until they read the console.
One webhook schema across four providers was the whole pitch. It held: our ingestion code has not changed since we integrated.
One flat plan, our own provider credentials, and the DND checks we were doing by hand. The maths was easy.
From your POST request to their pocket in under 500ms
Ingest
Edge gateways in Rust acknowledge in <50ms. Your request is durable before we return 202.
Watch every packet in flight
Overview
Throughput, delivery rate, and provider health at a glance.
Message Explorer
Trace any single message end-to-end, hop by hop.
Routing Rules
Author fallback chains and priority lanes without a deploy.
Journeys
Visual workflow builder for omnichannel campaigns. No code.
Compliance Center
DND hits, residency map, consent state — all auditable.
Analytics
Funnels and vendor latency, at scale. Sub-second.
Built for people who read changelogs
Strictly-typed SDKs, with Node available today and the rest of the set rolling out. Protobuf or JSON. A local CLI with environment detection, secret management, and webhook forwarding. An MCP endpoint so your AI coding agent can read schemas, write integration code, and run sandbox tests — safely.
- Node
- PythonCOMING SOON
- GoCOMING SOON
- JavaCOMING SOON
- PHPCOMING SOON
- .NETCOMING SOON
import { Nervly } from "@nervly/sdk"; const nervly = new Nervly({ apiKey: process.env.NERVLY_API_KEY }); await nervly.events.trigger( { name: "otp.requested", to: { subscriberId: "user_8x2", phone: "+2348011223344" }, payload: { code: "402913" }, }, { idempotencyKey: "ord_29441" }, ); // → 202 Accepted · { eventId, status: "QUEUED" }
Your questions, answered
Twilio is one provider. Nervly orchestrates the providers you already use — the ones shown above and more — with automatic failover, live cost arbitrage, and one normalized webhook schema. When a provider degrades, your messages don't.
Circuit breakers detect elevated error rates and reroute to the next healthy provider in your chain — automatically, in-region, without dropped messages. You'll see the failover in your console; your users won't see anything but the message.
Ingestion is an event-driven architecture: stateless Rust edge gateways buffer directly into NATS JetStream. We acknowledge in under 50ms at p99 and shape egress to each provider's real limits so nothing is dropped downstream.
Pre-flight checks against registries like Nigeria's NCC DND and India's TRAI run before dispatch. Blocked numbers are automatically rerouted to compliant transactional channels. GDPR and POPIA controls — residency, consent, erasure — are tenant-level configuration, not custom builds.
Where you tell it to. Data residency is configured per tenant: EU user data never leaves EU infrastructure. Internally, PII is replaced with format-preserving tokens, so raw phone numbers and emails never appear in logs or analytics.
No. This site sets no cookies, runs no third-party analytics, and writes nothing to browser storage — which is why there is no consent banner to dismiss. Client-side errors go only to our own first-party endpoint, with the URL query and hash stripped and no identifiers. It is a claim we enforce in the build, not just a line in a policy: the build fails if a cookie, a browser-storage call, an analytics vendor, or a third-party script origin is ever introduced.
It is one flat plan: ₦50,000 a month for 500,000 notifications and 100 requests a second, billed in NGN. You bring your own provider credentials, so carrier costs are billed by your provider at cost — we add nothing to them. Start free on 1,000 notifications a month.
Signal, delivered.
Your first 1,000 notifications are on us. Integrate in an afternoon.