Agents need a kill switch: when they run out of budget, go off the rails, or get stuck.
Restate gives you a full control pit: pause/resume, cancel (with rollback), kill, time travel, and restart on the same or updated code.
For one execution or a batch, via UI or in code.
Build durable agents & resilient backends with your coding agent.
Claude, Codex & Cursor now get Restate skills to design, build, debug & migrate apps.
MCP docs + TypeScript, Python, Java, Go support.
Let us know what you are building!
One of the things developers like most about Restate is how easy it is to deploy.
Restate is a single binary. To run it as a highly available cluster, deploy the same binary multiple times and snapshot to object storage.
No databases. No queues. No extra dependencies.
Yet you get:
• Durable Execution (retries & recovery)
• Virtual Objects (stateful entities)
• Durable RPC and messaging
Powered by our distributed replicated log, which provides fast replication, active-active durability, and flexible reconfiguration.
We just dropped Restate Java SDK 2.8.0 with new tracing support
Building Restate apps using #SpringBoot? Got ObservationRegistry already setup? Just bump the SDK and you get durable handlers tracing for free.
https://t.co/dxTI2TQE9l
Honored that @restatedev is included in @Redpoint's Infrared 100. We are in strong company.
It is the most intense and exciting time to build infrastructure for Agents and backends.
Happy to see that our take and durable execution resonates both with fast-growing startups and enterprises alike.
They picked Restate for its:
- Ease of self-hosting: critical for sensitive data and controlling costs
- Programming model: durable HTTP handlers with flexible control flow
- Serverless support with pause/resume while agents wait
Read the customer story:
https://t.co/0VexTgu7iJ
@Coralogix built an autonomous AI observability agent that answers questions across logs, traces, and alerts. Restate powers its:
→ Resilient semantic-layer pipelines: resume where they failed - no re-analyzing GBs of logs
→ Long-running background agents
As agents become more autonomous + long-running, reliability stops being a “nice to have” and becomes core product infrastructure. Aient is a great example.
Super cool work by @henrikfeldt 👏
Read more: https://t.co/LkIrOtkDYo
Aient is building a production-aware AI DevOps agent.
The agent harness is built on Restate for:
• Durable execution
• Resumable tools in remote sandboxes
• Per-agent inboxes + serialized turn-taking
• Human feedback via Slack/Linear
• Suspend/resume while idle
🧵Read more
New Pydantic AI integration with @restatedev.
Add durable execution to your agents in a few lines:
• Retries + journaled recovery
• Stateful sessions keyed by user
• Human-in-the-loop that survives crashes
• Multi-agent orchestration
Guest post with details: https://t.co/plSq0rcjTf
🚀 @restatedev is at an inflection point. We're hiring exceptional people to build the next foundational infra for AI apps & workflows. Remote, EU + US.
⚙️ Sr/Staff Cloud Infra
🛠️ Solutions / Field Eng
📣 DevRel Lead
🎨 Founding Brand Designer
👉 https://t.co/RRND62lTgS
Long-running agents need human approvals, feedback, sign-offs,... But long waits mean you need to be able to survive failures, often leading to complex event plumbing. Restate adds pause/resume to human approval steps for most popular agent SDKs and home-grown agent loops.
New integration: Restate + @langfuse for resilient, observable agents
Restate for retries, recovery, idempotency, and durable workflows. Langfuse for tracing, evals, and prompt versioning.
Together, you get a full stack for running agents reliably in production.
⬇️ Blog post