AI BriefingAnthropic Unveils Autonomous Coding Agent
Key point
Anthropic unveiled in London a coding agent that autonomously handles everything from writing PRs to deployment.
Details
Anthropic engineer Jeremy Hadfield demonstrated a new autonomous coding agent at the Code with Claude event held in London on May 19. Unveiled during this two-day event, which was held to coincide with the start of Google I/O, the system independently writes code, runs tests, and deploys pull requests without continuous human intervention.
The updated Claude model goes beyond simple autocomplete to independently handle entire codebases, and through its adaptive thinking feature, it determines on its own whether additional processing time is needed for complex tasks.
Benchmark performance:
- Industry-leading on Terminal-Bench 2.0
- 144 Elo points ahead of OpenAI GPT-5.2 on evaluations of economically valuable knowledge work
- 190 Elo point improvement over the previous version
Developers can assemble teams of agents for large-scale projects via the API. With the adoption of these autonomous systems, developers' focus is expected to shift from simple coding to platform architecture and planning.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
US Appeals Court Upholds Pentagon Blacklisting of Anthropic as Supply Chain Risk