Anthropic Expands Agentic AI Research Preview — Self-Improving Long-Duration Agents Now in Enterprise Beta

Anthropic launched a research preview of managed agents capable of handling long-running workflows autonomously in coding, finance, and law, alongside expanded public beta access to tools that allow agents to coordinate sub-agents and evaluate their own work using rubric-based outcome scoring. The initiative is framed as part of a broader vision for increasingly self-managing AI systems operating independently over extended periods. For AI product marketers working in or alongside these categories, this shifts the positioning conversation from "AI that assists tasks" to "AI that owns workflows end-to-end." The GTM implication is significant: product pages, sales decks, and competitive comparisons now need to address workflow ownership, accountability, error recovery, and human override mechanisms — the questions that will define enterprise evaluation criteria as agentic deployment becomes standard.

Agentic AIEnterprise GTMProduct LaunchAnthropic