GitHub Copilot, Cursor, Claude Code: AI Dev Tools Through the PO Lens

The explosion of AI dev tools: a PO's perspective

In 2025, developers in your teams are probably already using GitHub Copilot, Cursor, or Claude Code. As a Product Owner or IT Project Manager, you don't need to master these tools in depth — but you do need to understand what they change in your interactions with the team.

The three leading tools explained simply

GitHub Copilot: intelligent autocomplete

Copilot integrates directly into code editors (VS Code, JetBrains). It suggests code in real time, completes entire functions, generates unit tests, and answers questions in chat mode. It's the most widespread tool: 1.8 million subscribers in 2025.

What this changes for you: development estimates for repetitive tasks (CRUD, standard integrations, tests) should be reduced by 30–50%.

Cursor: the AI-first IDE

Cursor is a complete IDE built around AI. Its key feature: "Composer" allows simultaneously modifying multiple files based on a natural language instruction. It understands an entire codebase and can perform complex refactorings in seconds.

What this changes for you: refactorings and technical migrations, often very expensive to estimate, can be completed much more quickly. Revise your estimates for these categories.

Claude Code: the autonomous development agent

Claude Code (available in beta since early 2025) is a development agent that works in the command line. It can analyse a codebase, identify bugs, implement complete features, write tests, and commit its changes. It's the most autonomous of the three tools.

What this changes for you: for certain maintenance tasks or clear specification implementations, Claude Code can work near-autonomously. This changes the nature of what you expect from your developers.

Impact on the agile process