Claude Code Adds Projects, Letting Cloud Agents Run After You Log Off

Claude Code shifts to persistent, cloud-based multi-agent sessions

On September 17, Anthropic put Claude Code Projects into beta, reworking the tool around parallel, cloud-hosted threads instead of a single live terminal session. MarkTechPost and 富途牛牛 both describe tasks that keep executing after a developer closes their laptop, with a conversation split into separate threads that can run concurrently. VentureBeat frames the result as an 'always-on' workspace that remembers context and delegates long-running development work rather than depending on a user staying connected.

The Verge's description of the update — retooling Claude Code 'to manage multiple AI agents in the cloud' — points to Anthropic positioning the product less as a single autocomplete-style assistant and more as an orchestration layer for several agents working different parts of a codebase at once. The practical change for developers is that a task queued in Claude Code no longer depends on keeping a machine open; state-tracking and reliability move onto Anthropic's own infrastructure.

Sources: Claude Code Launches Projects: Splitting Conversations into Parallel Threads, with Tasks Continuing Even After Closing the Laptop — 富途牛牛 · Anthropic Launches Claude Code Projects in Beta: Parallel Cloud Sessions That Keep Running After You Close Your Laptop — MarkTechPost · Claude Code relaunches Projects to manage multiple AI agents in the cloud — The Verge · Anthropic launches Claude Code Projects, an 'always-on' conversation that remembers and delegates your long-running dev work — VentureBeat

A record pre-IPO valuation, and a warning about the reckoning after it

Business AM reported that Anthropic's pre-IPO valuation has reached approximately €1.73 trillion, a figure the outlet ties to the broader surge in AI industry valuations rather than to any specific new Anthropic disclosure that day.

Barron's used that same backdrop to argue that whenever Anthropic eventually lists publicly, the event will function as a stress test for AI-sector valuations as a whole — a moment where public markets, rather than private funding rounds, are left to price the technology's promises. The piece is framed as analysis of what a future listing would mean for the sector, not as a report of a confirmed IPO date.

Sources: Valorisation pré-introduction en bourse d'Anthropic atteint environ 1 730 milliards d'euros dans le sillage de l'essor de l'IA — Business AM · Anthropic's IPO Will Be AI's Next Moment of Crisis — Barron's

A beta opens for verified life-sciences users

Unite.AI reported that Anthropic has opened a Life Sciences Verification Program in beta, extending Claude's specialized offerings into a credentialed, domain-specific track alongside its general consumer and developer products. The day's coverage did not detail the mechanics of how users are verified or what capabilities the program unlocks.

Sources: Anthropic Launches Life Sciences Verification Program in Beta — Unite.AI

Rivals crowd the same terrain: verticals, agents, and monetization

The same day, OpenAI pushed into an adjacent vertical, launching what SiliconANGLE described as 'Astra for Law,' a GPT-6 configuration built for legal research; Reuters characterized the move as escalating the contest for law-firm customers. Fox Business reported that OpenAI also disclosed new incidents of 'concerning behavior' alongside the rollout of a new safety framework, with both announced together though the day's coverage did not detail the incidents' causes.

OpenAI was also reported to be testing conversational agents that speak with users on behalf of advertisers, per fr.qz.com. Set against Anthropic's Claude Code Projects launch and its life-sciences program, the pattern across the day's reporting is a market moving quickly from single-session assistants toward persistent, delegated, and increasingly monetized agents, with both leading labs pushing into specialized, credentialed, or commercial use cases at the same time.

Sources: OpenAI launches Astra for Law, a GPT-6 configuration for legal research — SiliconANGLE · OpenAI launches legal-focused AI platform, escalating race for law firm users — Reuters · OpenAI flags new concerning behavior incidents and launches safety framework — foxbusiness.com · OpenAI teste des agents d'IA qui discutent avec les utilisateurs au nom des annonceurs. — fr.qz.com