DigitalOcean opened Managed Agents in public preview, bundling isolated, persistent harness runtimes, governed access to 16,000+ tools, and inference across 75+ open and proprietary models. It supports familiar environments including Claude Code, Codex CLI, OpenCode, LangChain, and LangGraph, with MCP-compatible tool access; OpenHands, Qencode, and Amplitude are already building on it. The notable shift is infrastructure vendors packaging the agent runtime/tool/security/inference stack as one cloud primitive rather than leaving teams to assemble it. Official launch claims; independent performance and security comparisons remain to be seen. Source: https://investors.digitalocean.com/news/news-details/2026/DigitalOcean-Launches-Managed-Agents-Bringing-Agent-Execution-Tool-Access-and-Inference-Together-on-One-Cloud/default.aspx
Pendo launched Pendo for Agents, a product-analytics/context offering intended to supply agents with live product information and let teams measure agent adoption and outcomes. This is a less obvious agent-infrastructure angle: it treats software-product telemetry and user context as a runtime input and evaluation surface for agents, not merely an analytics dashboard. Company-described capabilities; independent evidence of impact is not yet available. Source: https://www.pendo.io/news/pendo-launches-pendo-for-agents/