DigitalOcean adds services for self-improving AI agents

what happened

BusinessWire reports that DigitalOcean has introduced Agent Learning Services and Agent Data Services to help teams build AI agents that improve through repeated use. The offerings extend the company’s Inference Engine and Managed Agents, creating a single platform for deploying, operating and refining agents.

Agent Learning Services include Insights, Signals, and Evaluations & Simulations. The tools capture an agent’s actions, identify failures, and test proposed changes against realistic conversations before customers encounter them. The Inference Router optimizer selects a model for each task according to cost, quality, or a customer’s own evaluations. DigitalOcean plans to apply the same automated loop to agent workflows, alongside existing human approvals.

Agent Data Services provides a managed data layer where video is transcribed and described, documents are indexed, and logs and agent sessions are assembled with configurable redaction for sensitive information. Agent Learning Services are in public preview, while the Inference Router optimizer and Agent Data Services are in private preview. DigitalOcean will show the products on Oct. 13 in San Francisco at its Open Intelligence Summit. The company says more than 680,000 customers and millions of developers use its services.

why it matters

The announcement broadens DigitalOcean’s AI offering beyond infrastructure into data and learning tools. Near-term revenue impact is difficult to assess because the services are previews, are not supported for production workloads, and may be changed, withdrawn, or never generally released.

instruments