AI and software development: the developer's craft re-tooled
Development is one of AI's most mature uses: generating code, reviewing it, writing tests, migrating legacy. This dossier tracks the transformation of the developer's craft, beyond mere autocomplete.
Latest AI & development news
- Claude Code 2.1.16 — [IDE] Fixed a race condition on Windows where the Claude Code sidebar view container would not appear on start — Claude Code
- Claude Code: Best practices for agentic coding — Anthropic (ingénierie)
- Claude Code 2.1.169 — Fixed enterprise managed MCP policies (allowedMcpServers/deniedMcpServers) not being enforced on reconnect, IDE-typed configs, - — Claude Code
- Ask HN: Multi-LLM orchestration frameworks that collaborate? — Hacker News
- Show HN: Neal – Codex writes the code, Claude reviews it — Hacker News
- Claude Code vs OpenAI Codex: What Published Coding Tests Actually Show — quasa.io
- Show HN: Finterm.ai Bloomberg terminal for Claude Code — Hacker News
- Show HN: Termic – desktop app for running CLI coding agents (Claude Code, Codex) — Hacker News
- Show HN: AgentCodeGUI – Multi-Account Desktop GUI for Claude Code and Codex — Hacker News
- Show HN: Ski – Voice Coding for Claude Code, Codex and More – On-Device – Free — Hacker News
From autocomplete to full assistant
AI has moved from suggesting a single line to assistants that can read a whole repository, propose an implementation, explain a bug and run tests. Claude Code illustrates this shift: AI no longer just completes, it executes development tasks under supervision.
Review, tests and quality
Beyond writing, AI helps review code, spot flaws, generate tests and document. This is often where the value is safest: a fast second look that catches what a rushed human misses, while never replacing the final human review.
Legacy, migration and technical debt
Understanding old code, translating it to another language or framework, documenting a forgotten codebase: AI excels at reconstructing lost context. It speeds up migration projects, provided every change is validated by tests.
Frequently asked questions
Will AI replace developers?
No: it automates tasks (generation, review, tests) and boosts productivity, but design, architecture and responsibility for the code remain human.
Can AI-generated code be trusted?
Not blindly: code must be reviewed and tested. AI can introduce subtle bugs or flaws; human review remains essential.
Is Claude good at code?
Yes, Claude is regularly on top for agentic coding; Claude Code lets it read, edit and test a project. See our comparisons.
Claude News is published by Héra SASU. Independent media, not affiliated with Anthropic.