Summary

May 19 centered on infrastructure maturing around production agents: Anthropic pushed managed agents closer to customer-controlled environments, while Vercel added more routing, security, and operational controls across inference and deployment workflows. Alongside those platform moves, newer products like Ludr and Fere AI showed agents spreading into more specialized execution surfaces, from screen interaction to trading.

Key themes

  • Enterprise agents are getting closer to private infrastructure, with stronger sandboxing and private tool connectivity becoming part of the managed-agent story.
  • Inference and developer platforms are adding policy, routing, and security controls, especially around multi-provider model traffic and protected production workflows.
  • New agent products are pushing beyond chat into execution surfaces such as on-screen assistance and autonomous trading actions.

Notable items

  • Anthropic added self-hosted sandboxes and private MCP tunnels to Claude Managed Agents, strengthening the case for enterprise deployment with tighter runtime control.
  • Vercel expanded AI Gateway with cost-, latency-, and throughput-based provider sorting plus routing metadata, signaling a shift toward policy-driven inference control planes.
  • Vercel also shipped developer workflow and security updates through authenticated vercel curl access to protected deployments and Protected Source Maps.
  • Ludr launched a screen-aware assistant for macOS that turns visible UI context into explanations, edits, and coding-agent handoff.
  • Fere AI highlighted a specialized agent model for turning market signals into crypto and prediction-market trades.

Source coverage

Source rows used: 10