The Complete Guide to “Code with Claude” 2026: Everything Anthropic Just Announced

Anthropic just wrapped up its second annual developer conference, Code with Claude, in San Francisco on May 6, 2026. The event was a watershed moment for the AI industry, marking the transition from “chatbots” to fully autonomous, self-correcting agents.

Here is the complete recap of every major announcement, from the massive SpaceX partnership to the new “dreaming” capability for AI agents.

The Event: A Brief History

While Anthropic has become a household name in AI, its developer conference is a relatively new tradition. The inaugural event took place in 2025, focusing on the launch of the Model Context Protocol (MCP). In just one year, the conference has evolved from an intimate gathering into a global tour. This year’s San Francisco summit serves as the flagship kickoff, with subsequent events scheduled for London (May 19) and Tokyo (June 10).

Who is Anthropic?

For those new to the ecosystem, Anthropic is an AI safety and research company based in San Francisco. Founded by former OpenAI executives, the company is defined by its mission to build “reliable, interpretable, and steerable” AI systems. They are best known for Claude, their family of large language models, and “Constitutional AI,” a training method that aligns models with a set of safety principles. In early 2026, CEO Dario Amodei revealed the company experienced an unprecedented 80-fold growth, cementing its place as a primary rival to the world’s largest tech giants.

1. The Infrastructure Moonshot: Partnership with SpaceX

To handle its explosive growth, Anthropic announced a massive compute partnership with SpaceX.

• Project Colossus: Anthropic will utilize SpaceX’s Colossus 1 data center in Memphis, gaining access to over 220,000 Nvidia GPUs and 300 megawatts of power.

• Orbital Compute: In a sci-fi twist, the companies are exploring a partnership to develop gigawatts of orbital AI compute capacity, literally putting data centers in space to bypass terrestrial power constraints.

2. The Next Generation of Agents: “Dreaming” & Self-Correction

The technical highlight of the event was a suite of upgrades for Claude Managed Agents, designed to make them autonomous and reliable enough for real production work.

• Dreaming (Research Preview): This feature allows agents to review their own logs and past work during “downtime” (between sessions). By spotting patterns and self-correcting errors offline, agents improve their performance without human intervention.

• Outcomes (Public Beta): Developers can now define a “rubric” for success. A separate “grader” model evaluates the agent’s work against this rubric, forcing the agent to retry and self-correct until the output meets the quality standard.

• Multi-Agent Orchestration: A new “lead” agent can now break down complex tasks and delegate them to specialized sub-agents (e.g., one for coding, one for research) that work in parallel.

3. Enterprise & Finance Solutions

Anthropic is moving deep into vertical integration with tools specifically for the financial sector.

• Financial Agent Templates: The company released 10 ready-to-run agent templates for tasks like KYC screening, pitchbook generation, and month-end closing.

• Microsoft 365 Add-ins: Claude is now integrated directly into Excel, PowerPoint, and Word, with Outlook coming soon. This allows context to carry over between apps, so an analysis in Excel can be instantly turned into a slide deck in PowerPoint.

• New Data Connectors: Secure integrations were announced with major data providers, including Moody’s, Dun & Bradstreet, and Fiscal AI.

4. Developer Tools & Security

For the builders in the room, Anthropic rolled out significant quality-of-life improvements.

• Claude Security (Public Beta): A new tool for Enterprise customers that scans code repositories for vulnerabilities and proposes fixes using the powerful Opus 4.7 model.

• Claude Code CLI Updates:
  • Doubled Rate Limits: Usage limits for Pro, Max, Team, and Enterprise plans have been doubled.
  • No More Peak Caps: Peak-hour throttling has been removed for Pro and Max accounts.
  • Project Purge: A new claude project purge command allows developers to wipe all state and history for a specific project.

The Bottom Line: From Chatbot to Co-Worker

The 2026 San Francisco summit made one thing abundantly clear: Anthropic is no longer interested in just having the “smartest” model in the room. They are building an industrial-grade ecosystem designed for action.

By solving the two biggest hurdles for AI, compute scale (via the SpaceX deal) and reliability (via the “Dreaming” and Outcomes features), Anthropic is signaling that the era of the experimental chatbot is over. We have entered the era of the autonomous AI workforce. For developers, the message is simple: the tools are finally powerful enough to handle the “boring” stuff, so you can focus on the big ideas.

As the conference moves on to London and Tokyo, the eyes of the tech world remain on San Francisco. If these announcements are any indication, the way we build software and manage data is about to change forever.