World's Fair 2025
Collaborating with Agents in your Software Dev Workflow - Jon Peck & Christopher Harrison, Microsoft
Jon Peck , Christopher Harrison
Overview
This talk explores how developers can effectively collaborate with AI agents, specifically focusing on GitHub Copilot's capabilities within the software development workflow. The core thesis is that understanding and providing proper context is crucial for maximizing the AI pair programmer's utility, moving beyond simple prompt engineering to encompass code readability, project structure, and clear intent. The discussion highlights various Copilot features, from basic code completion to advanced agent modes, and emphasizes best practices for integrating these tools into daily development tasks.
Who should watch
- AI engineers and developers looking to integrate AI assistants into their workflow.
- Product Managers and technical leads evaluating AI tools for team productivity.
- Builders and developers interested in understanding the practical application of AI in software development.
- Anyone struggling with prompt engineering and seeking to improve AI interaction through better context.
Key takeaways
- Context is paramount for AI pair programmers like GitHub Copilot, extending beyond prompts to include code readability, clear naming conventions, comments, and project structure.
- GitHub Copilot offers several modes of interaction: code completion for immediate suggestions, chat for single-turn queries, edits for multi-file modifications, agent mode for iterative coding within the IDE, and coding agent for automated task completion on GitHub.
- Effective collaboration with Copilot involves clearly stating intent and avoiding terse prompts, treating the AI as a partner that needs explicit information.
- Copilot Coding Agent operates within GitHub.com, can be assigned issues, and works iteratively, running tasks like unit tests and self-healing.
- Security is a key consideration for Copilot Coding Agent, which by default has no internet access, runs in ephemeral environments, and has restricted write access to repositories, only modifying its own branch.
- To improve AI performance and consistency, developers can leverage Copilot instructions files and Model Context Protocol (MCP) servers to provide additional context and enable access to external data.
- AI tools do not replace the established DevOps flow; manual code reviews, linters, security scans, and unit tests remain essential safeguards.
- When AI agents make mistakes, strategies include clarifying intent via comments, unassigning and reassigning the issue, or refining the context provided in instructions files.
Notable quotes
*Context is key.*
*Don't be passive-aggressive with Copilot.*
*AI does not change the DevOps flow.*
Unofficial community note. Prefer the recording for nuance.