AI for software development
Using language models and coding agents to write, refactor, test and review code.
- Commonly used tools
- Claude Code, Codex CLI, Cursor, GitHub Copilot
Agentic tools perform best with standing instructions (CLAUDE.md or AGENTS.md), exact test commands and a human reviewing diffs. Controlled studies disagree on speed gains, so measure accepted work rather than generated lines.
Connected
- Claude CodeAnthropic's agentic coding tool that reads a codebase, edits files, runs commands and verifies its work from the terminal, IDE or web.
- Codex CLIOpenAI's open-source coding agent that runs in the terminal, released in April 2025.
- CursorAn AI-first code editor built on VS Code by Anysphere, with agent mode, codebase indexing and project rules.
- GitHub CopilotGitHub's AI pair programmer, launched in 2021, now including chat, code review and a cloud coding agent.
More tasks: what ai is used for
- AI for researchUsing AI systems to search, read and synthesise sources into cited answers.
- AI for writing and editingUsing language models to draft, restructure and edit text.
- AI for image generationCreating and editing images from text prompts.
- AI for speech and voiceConverting speech to text, text to speech and cloning voices.
- AI search visibility (AEO / GEO)Making a business understandable and citable by AI answer engines.
- AI for customer supportAssistants that answer customer questions and help human agents resolve tickets.