The agent-first cloud: why we built Managed Agents

DigitalOcean has introduced Managed Agents in public preview as an integrated platform for AI agents that combines isolated execution sandboxes, model inference, persistent data, tool access, and observability in one runtime. The company argues that agents differ from conventional VM-based workloads because they operate in short compute bursts, consume tokens, wait for model or human responses, and require durable state across sessions. Managed Agents uses Firecracker microVMs for isolated, resumable sessions, an Action Gateway for governed access to more than 16,000 tools while brokering credentials outside agent environments, and active-use billing intended to charge for CPU work rather than idle time. DigitalOcean positions the service as an alternative to multi-vendor stacks, emphasizing reduced latency, centralized security, unified telemetry, support for open-weight and frontier models, and compatibility with common agent frameworks and custom OCI images. The platform targets companies building agent products, teams automating operational work, and organizations running coding agents, while planned features including Signals, evaluations, Simulations, and an Iceberg-based data platform aim to help agents learn from production outcomes; however, DigitalOcean notes that preview capabilities and roadmap plans are subject to change.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.