AI BriefingRivet Launches Open-Source Actor Platform for Scaling AI Agents on S3
·2026.11.06 02:17
Key point
Rivet offers a lightweight, S3-based orchestration layer that achieves 4.99 ms p99 cold starts and 72 KB memory per actor, positioning itself as a scalable alternative to heavy sandboxes and Cloudflare Durable Objects.
Details
- Rivet is an open-source (Apache 2.0) orchestration platform designed to scale AI agents using lightweight Rivet Actors instead of heavy, isolated sandboxes.
- Performance benchmarks show Rivet Actors achieve a p99 cold start of 4.99 ms and use only 72 KB of memory per actor, compared to 1.70 s and 512 MiB for standard sandboxes.
- The architecture relies solely on S3-compatible object storage for persistence, eliminating dependencies on Postgres, FoundationDB, or NATS, and claims to be approximately 10.4x cheaper than replicated SSD storage.
- The platform supports Node.js, Bun, and Rust runtimes without custom V8 modifications, and offers deployment options including self-hosting, Rivet Cloud, and Bring Your Own Cloud (BYOC) within private VPCs.
- Key features include durable execution, real-time WebSocket connections, cron scheduling, and an agentOS component for executing untrusted code with fine-grained resource controls.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Bernstein – open-source governance layer for AI agentsINIT Labs releases INIT, a Rust agent harness that hot-swaps its own core mid-task
Microsoft Engineers Detail Copilot Cowork's 4.5-Month Path to GA and Python-to-Rust Migration