Humans set direction. Agents close the loop.
Mario Rodriguez and Evan Boyle frame agent value in terms of software work completed, rather than code produced. Their GitHub Universe 2026 session connects that idea to an engineering model built around human direction and agent execution.
From code generation to completed work
The session explores a development loop in which developers define outcomes and standards, while GitHub Copilot agents carry tasks through planning, review, security, and deployment. People remain accountable for what ships.
Connecting context, workflows, and enterprise controls is central to this model: teams need those foundations to delegate ambitious tasks with confidence.
Session topics and chapters
The presentation moves from collaboration and human direction to an agent-native engineering system, the GitHub Copilot app, community remixing, and automation:
- 00:00 — Welcome to GitHub Universe 2026
- 00:33 — Humanity’s progress through collaboration
- 02:42 — Humans set direction, agents close the loop
- 05:25 — The agent-native engineering system
- 06:48 — GitHub Copilot app and community remixing
- 16:08 — Raising the ceiling with GitHub Copilot agents
- 19:30 — Automating CI with lever
- 24:07 — Lever autonomy and cloud automations
- 27:30 — Reclaiming time for creative work
- 28:53 — Sharing work and imagining tomorrow
The central distinction is between delegating execution and retaining responsibility: agents carry work forward, while developers set its direction, define acceptable results, and stay accountable for delivery.