Run your agents in production. We'll handle the rest.
Swirls Cloud is the managed runtime for your .swirls agents. Same engine you run locally, run at production scale by us. Auto-scaling workers, the Portal, OpenTelemetry, and a verifiable audit chain are bundled in. Deploy with git push or swirls cloud deploy. No CI/CD to maintain.
Same engine. Production scale. Governance built in.
Cloud runs the same Swirls engine you develop against locally. Auto-scaling, observability, and cryptographic governance are bundled in. You describe the system. We run it.
Runtime, managed
- Durable execution with automatic step recovery
- Auto-scaling workers based on queue depth
- The Portal hosted: replay, inspect, approve
- Sleep, signal, and child-graph yields
Governance, enforced
- Scoped delegation tokens, attenuated per call
- Per-node secrets, never ambient
- Tamper-evident audit chain, customer-verifiable
- Permissions bound to the workflow hash
Regulated production
- 99.9% uptime SLA on every Cloud account
- Row-level tenant isolation, per-tenant keys
- OpenTelemetry tracing for every execution
- SOC 2 Type II targeting 2027
Three steps from local to production.
The development workflow is the production workflow. Same artifact, same engine, same security model. Local is the rehearsal. Cloud is the show.
Author your agent
Compose nodes across .swirls files. Declare triggers, schemas, OIDC auth, and the graphs each node can call as tools. The definition is the contract.
Run it locally
The full engine on your machine. Local-backed state, the Portal, the LSP. What runs locally runs in Cloud.
Ship it
No CI/CD pipeline to build. Push to your branch and we deploy on merge, or run swirls cloud deploy from your terminal. Either way the compiled definition gets locked to every execution token.
Everything your agents need to run.
Every Cloud deployment ships with the full platform. Observability, secrets, audit logging, human review, replay. No add-ons, no plugins.
Durable execution
The same engine you run locally, running in production. Steps are checkpointed automatically. Failovers without duplicate work.
Auto-scaling workers
Workers scale up and down with your queue depth. You pay for the work that ran, not the machines you reserved.
The Portal
Inspect every execution. Replay node-by-node. Approve human-in-the-loop reviews. Bundled with every deployment.
OpenTelemetry, built in
Every execution emits OTel spans. Every audit event lands in an append-only log you can verify.
Per-node secrets
Secrets are delivered to nodes at run time. A node only ever sees what the .swirls file declared it needs. There is no ambient credential context.
Team collaboration
Invite teammates, manage roles, and review each other’s deploys. Shared Portal, shared audit trail.
Uptime SLA
99.9% uptime SLA on every Cloud account. Enterprise plans run on dedicated infrastructure.
Versioned deploys
Every deploy is a snapshot. Roll back to any previous version with one command.
The .swirls file is the security policy.
Cloud deployments derive every permission, every key, and every token from the .swirls files you wrote. Modify a file and old tokens stop working. Encryption, scoping, and verifiability are structural, enforced by the runtime.
Build local for free. Priced for production.
Local is free forever. Developer is $49 a month, production Swirls for solo builders. Team is $799 a month for engineering organizations with SSO, role-based access, and longer audit retention. Every paid plan includes unlimited seats. Enterprise is custom for regulated deployments with BYO-KMS, on-prem, and attestation.
Five minutes from install to an agent you'd ship.
Install the CLI. Write .swirls files. Run them locally with the same engine that runs in production. When you're ready, git push or swirls cloud deploy. No CI/CD to build.
Building something with Swirls? Come hang out in Discord.
Join the Discord