Anthropic’s Code with Claude showed off coding’s future—whether you like it or not
Anthropic's "Code with Claude" event showcased a future where AI, specifically Claude, autonomously handles most coding tasks, with developers relying heavily on its code generation capabilities. This rapid shift, while embraced by many at the event, raises concerns among some coders about decreased human oversight, potential security risks, and the impact on developer skills, despite Anthropic's emphasis on existing best practices.
Anthropic's "Code with Claude" event in London highlighted the increasing adoption of AI in software development. Many attendees, when polled, indicated they had recently shipped pull requests almost entirely written by Claude, with a significant number admitting they hadn't even reviewed the AI-generated code. This illustrates a rapid shift in the industry, where LLM-powered tools are fundamentally changing how software is created.
This paradigm shift is happening quickly. While Claude could code somewhat last year, recent updates, particularly Claude 4.6 and 4.7, have made it a tool developers are increasingly comfortable entrusting with their work. Anthropic aims for ultimate automation, envisioning Claude as not only generating code but also correcting its own errors, ideally without human intervention. The company refers to this as "Claude prompting itself" and encourages developers to "let it cook."
Claude Code also features "dreaming," a new system where agents record notes on tasks, allowing subsequent agents to learn from past errors and expedite their work. This system consolidates information across tasks, helping Claude learn and improve its understanding of a codebase over time. This approach, though innovative, has generated mixed reactions within the wider developer community.
While the event itself showed enthusiasm, with no signs of unease among attendees, external reports reveal growing concerns. Some coders lament online that AI tools, driven by management's quest for productivity, make development harder due to the increased need to review generated code. Others worry about a decline in their own coding skills and the potential for AI to produce insecure code.
Anthropic acknowledges these concerns. Katelyn Lesse, an engineering lead, stressed that traditional software development best practices remain crucial, although some teams might be overlooking them. Angela Jiang, a product lead, added that the ambition is for Claude to eventually "build itself." Currently, Claude is considered comparable to a mid-level engineer, meaning expert human engineers are still necessary for system design and complex problem-solving, though the aim is for Claude to continuously improve across all engineering aspects.
Related articles
Build real agentic apps using CUGA: two dozen working examples on a lightweight harness
CUGA, IBM's open-source Agent Harness, simplifies building agentic applications by handling infrastructure, allowing developers to focus on tools and prompts. It offers pre-assembled components for planning, execution, and state management, significantly reducing development time. CUGA has topped agent benchmarks like AppWorld and WebArena.
OpenAI launches new initiative to help find and patch open source bugs
OpenAI has launched "Patch the Planet," a new initiative in partnership with cybersecurity firm Trail of Bits, to enhance the security of open-source projects. This program aims to assist maintainers in identifying and patching bugs, utilizing OpenAI's AI-powered security tools while reducing the burden on project teams.
PP-OCRv6 on Hugging Face: 50-Language OCR from 1.5M to 34.5M Parameters
Baidu has released PP-OCRv6, an advanced optical character recognition (OCR) model supporting 50 languages. Available on Hugging Face, this version significantly improves accuracy and efficiency across various parameter sizes, from 1.5 million to 34.5 million, marking a substantial leap in multilingual OCR technology.