The AI-native backend,
from laptop to fleet
One single surface and one AI-first CLI for your entire backend — identity, data, storage, functions, jobs, sites, AI gateway, and more. Run a single local node or a fully component-isolated fleet, and migrate between them without rewriting your project.
Think Supabase + Vercel + Segment + Airbyte + OpenRouter + Step Functions — unified behind one coherent surface. A robust Rust layer of adapters over best-in-class building blocks, so you never reinvent the wheel.
One surface, every primitive
Each capability is exposed through the same consistent, AI-first API and CLI — and they compose with one another.
Identity →
Auth, sessions, API keys, and multi-tenant orgs/teams with roles and permissions built in.
Data →
Relational data and instant APIs over PostgreSQL, with row-level security and smart caching.
Realtime →
Live subscriptions and presence on top of your data — push changes to clients as they happen.
Storage →
Object storage with signed URLs and CDN delivery. S3-compatible or local filesystem.
Functions →
Serverless compute with an adaptable runtime — WASM today, Lambda and Firecracker as you scale.
Jobs →
Scheduled and event-driven workflows with durable execution, retries, and dead-letter queues.
Sites →
Static and dynamic site hosting at the edge, with custom domains and automatic TLS.
AI Gateway →
LLM routing with per-user quotas, residency, and cost controls — the Relai concept, built in.
Connect →
Sync jobs, ETLs, and API calls — a Connector trait composed from Functions, Jobs, Data, and Storage.
Analytics →
Product and usage analytics with an event pipeline — Segment-class, on your own infrastructure.
Plus Secrets, Policies, and Cache — all behind the same surface.
Self-managing, self-healing
Reactor can operate itself. It detects and diagnoses failing components, generates a new version, sandbox-tests it, and — with your authorization — promotes it to production. Full AIDevOps, under human supervision.
Detect
Continuous health checks surface failing or degraded components.
Diagnose
Reactor reasons over telemetry to pinpoint the root cause.
Generate
A new component version is generated to fix the issue.
Sandbox test
The candidate is validated in isolation before it ships.
Promote
With human authorization, the fix is promoted to production.
From one node to a fleet
A Reactor cluster scales — and migrates — across six topologies. The Rust abstraction layer makes moving between sizes possible without rewriting your project, in any cloud.
Local
Single node
PG isolated
Infra isolated
Gateway isolated
Fully isolated
Self-host or cloud
Run the open-source core anywhere, or let Reactor Cloud manage the fleet for you.
Open source
reactor · Apache-2.0
- Single binary, runs anywhere
- Your infrastructure, your data
- Server, CLI, and console included
- Docker, Fly.io, AWS — any cloud
Reactor Cloud
Managed fleet — free to start
- Free projects in a shared cluster
- Usage-based paid upgrade paths
- Provisioning, billing, and migration
- Enterprise self-host licensing
Ready to build?
Spin up your first project in minutes with the quickstart guide.