Boost Productivity with Copilot in Visual Studio

Key Concepts

• Agent Preview: A new harness in Visual Studio that integrates with the Copilot CLI SDK, enabling advanced automation features.
• Plan Mode: A feature that allows Copilot to generate comprehensive, step-by-step implementation plans (saved as Markdown files) based on user requirements.
• Work IQ: A tool that extracts requirements and action items directly from meeting transcripts (e.g., Microsoft Teams) to generate development plans.
• Test Agent: An automated tool that analyzes code coverage gaps and generates unit tests at the file, project, or solution level.
• Model Picker: A configuration interface allowing users to select specific LLMs (e.g., GPT-5) and view cost/credit implications.
• Bypass/Autopilot Mode: Modes that allow Copilot to execute tasks with varying levels of human oversight.

---

1. Main Topics and Workflow Enhancements

Wendy Brighting, a Product Manager on the .NET team, demonstrated upcoming features for Visual Studio 2026 (scheduled for a June release). The core focus is on asynchronous, multi-agent workflows that allow developers to handle planning, coding, testing, and reviewing simultaneously.

• Parallel Processing: Users can run multiple Copilot sessions concurrently (e.g., code reviews, test generation, and planning) without blocking the IDE.
• Session Manager: A centralized dashboard to track the status of all active Copilot tasks, allowing users to monitor progress or intervene if a process "goes off the rails."
• Customizable Skills: Developers can define specific "skills" (via Markdown files) that dictate how Copilot performs tasks like code reviews, ensuring the AI follows specific organizational standards or personal preferences.

2. Step-by-Step Methodologies

Code Review Framework

1. Skill Definition: The user defines a custom skill that mandates a parallel analysis using two different models (e.g., a cloud model and a GPT model).
2. Execution: Copilot performs a git diff analysis, runs existing tests, and checks for security/thread-safety issues.
3. Synthesis: The AI reconciles findings from both models and presents a structured report categorized by "Critical Fixes," "Suggestions," and "Passed Checks."

Test Generation Process

1. Scope Definition: The user specifies the scope (e.g., a specific project or file).
2. Gap Analysis: The agent runs existing tests to perform a code coverage analysis.
3. Generation: The agent identifies uncovered logic and generates unit tests, followed by a build and test run to ensure the new code is functional.

3. Real-World Applications

• Meeting-to-Code Pipeline: Using Work IQ, the user pulls transcripts from multi-session meetings to automatically generate a technical specification and implementation plan, eliminating the manual effort of writing documentation.
• Modernization: The Modernize Agent assists in upgrading legacy applications (e.g., .NET Framework to .NET Core) by providing a step-by-step migration plan that can be executed incrementally.
• Performance Optimization: The Profiler Agent simplifies the complex task of performance tuning by identifying bottlenecks and suggesting specific code changes.

4. Key Arguments and Perspectives

• Efficiency through Delegation: Brighting argues that by offloading repetitive tasks (like writing unit tests or summarizing meetings) to AI, developers can focus on high-level feature design.
• Human-in-the-Loop: Despite the power of "Autopilot" modes, Brighting emphasizes the importance of being "token conscious" and maintaining oversight, noting that she prefers to watch the agent's progress to ensure it stays aligned with project goals.
• Quality Assurance: By performing multi-model code reviews before submitting code to the team, developers can reduce the burden on their peers and ensure higher-quality code reaches the repository.

5. Notable Quotes

• "I don't want to just sit here and watch it... that's why we have these options. It's amazing." — Regarding the ability to run multiple asynchronous Copilot sessions.
• "Who likes to write unit tests? We want to create new features, not write tests, and this now does it for us." — On the value proposition of the Test Agent.
• "This particular Work IQ tool has been a lifesaver for me... I no longer have to [manually translate conversations into a spec]."

6. Synthesis/Conclusion

The upcoming Visual Studio 2026 updates represent a shift toward agentic development, where the IDE acts as a project manager, tester, and reviewer. By integrating meeting transcripts (Work IQ), multi-model code analysis, and automated test generation, Microsoft is aiming to reduce the "administrative" overhead of software engineering. The ability to run these tasks in the background—even on a Dev Box while the laptop is closed—significantly optimizes the developer's workflow, allowing for a more continuous and productive coding cycle.