From Skeptic to Superpower: Real‑World AI Coding Workflows That Scale | BRK229
Mario Toffia and Priyanka Sharma share a practical look at AI-assisted coding workflows, comparing Claude Code + Cursor with GitHub Copilot CLI and focusing on what works, what breaks down, and how teams can scale usage without losing control over sensitive infrastructure.
Overview
This Build 2026 breakout session compares two day-to-day AI coding setups used by engineers:
- Claude Code + Cursor
- GitHub Copilot CLI
The speakers focus on realistic outcomes rather than hype: initial skepticism, where each tool is strong, where it fails, and what changes when a team scales AI-assisted development to larger delivery expectations.
Key themes covered
Background and context
- The session opens with speaker background context (cloud native, Kubernetes, telco/IoT, and early LLM experimentation).
Autonomous/agentic coding workflow with Copilot CLI
- The session includes a segment where an autonomous coding workflow is initiated using GitHub Copilot CLI.
Where AI helps (and what it changes)
- Discussion includes how generative AI can help backend engineers produce functional UI output.
- The speakers frame this as “new tech, old problems,” emphasizing that familiar engineering constraints still apply.
Reliability and operational risk: “AI collaborators can quit anytime”
- A key lesson highlighted is that AI tools can be inconsistent or fail unexpectedly, so teams need processes that assume the assistant may stop being helpful at any time.
Maintaining manual control for sensitive infrastructure
- The session stresses keeping human control for high-risk areas (especially sensitive infrastructure work), rather than delegating end-to-end changes blindly.
Scaling to a team setting
- The talk describes what these workflows look like when used by a team (described as 20 engineers delivering at a scale of 200), including trade-offs and lessons learned.
Session metadata
- Event: Microsoft Build 2026
- Session code: BRK229
- Level: Intermediate
- Speakers: Mario Toffia, Priyanka Sharma
- Resource link: https://aka.ms/build26-next-steps