| status | active | |||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| updated | 2026-10-30 | |||||||||||||||||||||||||||||||||||||||||||||||||
| section | Foundations | |||||||||||||||||||||||||||||||||||||||||||||||||
| audience |
|
|||||||||||||||||||||||||||||||||||||||||||||||||
| level | foundational | |||||||||||||||||||||||||||||||||||||||||||||||||
| duration | 45 | |||||||||||||||||||||||||||||||||||||||||||||||||
| format | core-talk | |||||||||||||||||||||||||||||||||||||||||||||||||
| decision | Which Copilot surface fits this work, context, and reviewer? | |||||||||||||||||||||||||||||||||||||||||||||||||
| prerequisites | ||||||||||||||||||||||||||||||||||||||||||||||||||
| related |
|
|||||||||||||||||||||||||||||||||||||||||||||||||
| references |
|
The Question This Talk Answers: "Which Copilot surface fits this work, its constraints, and its desired outcome — and what capability does the team build next?"
Duration: 45 minutes | Audience: Cross-functional teams choosing where agentic work happens
Thesis: Start with eligible surfaces, preserve the work's decisive context, then match control, review ownership, and delivery model.
- Open the deck: 28 animated slides. Press
Nto open the speaker notes beside it. - Speaker notes: the talk track, facts for questions, first-party sources, and a time budget for every slide. The notes are the reference for this talk; they print as a handout.
The deck and notes live in slides/tech-talks/surfaces/slides.js. Edit talk content there; this page stays a short landing page.
Seven Copilot experiences on two axes, where you steer and where the work runs, plus a routing tree from desired outcome to decisive context.
Each role gets one useful first task, one reviewable artifact, and one clear handoff to the next surface.
Four phases (activate, standardize, delegate, orchestrate), each crossed with reviewed evidence from normal work and tracked on a per-workflow scorecard.
A production incident routed across the CLI, VS Code, and the coding agent, plus three handoff patterns and five boundary signals for changing surfaces.
- Try: Pick one task in flight and name its decisive context, its delivery artifact, and the reviewer who accepts it.
- Expected signal: The three answers point at one surface from the routing tree.
- Validate: Ask the named reviewer whether that artifact is what they would accept.
- Build: Classify three recent tasks with
surface-decision-tree.mmdand add your team's constraints tosurface-contracts.yml. - Expected signal: One documented handoff trigger, such as "the issue is ready to implement".
- Validate: Compare the router's picks with the surfaces the team actually chose, and record where they differ.
- Pilot: Baseline the team across the four phases and practice all three lifecycle patterns on normal work.
- Success measure: Local thresholds set before advancing a phase: minimum sample size, acceptance signal, quality guardrail, and escalation limit.
- Boundary: When a boundary signal fires, change surfaces at the transition and name the artifact, evidence, reviewer, and escalation trigger.
- Candidate task: A piece of work your team routes by habit today.
- Decisive context: Where its strongest context lives: open files, a terminal session, an issue, or a product runtime.
- Delegation and authority: Which surface does the work, what stays human-led, and who approves the delivery artifact.
- Evidence: The reviewed artifact, and whether the reviewer accepted it without re-routing.
surface-contracts.yml: comparable contracts for all seven Copilot experiences.surface-decision-tree.mmd: Mermaid source for the routing flowchart.role-goal-scenarios.yml: test cases showing how roles, goals, and constraints resolve to a surface.adoption-roadmap.yml: training paths from first engagement through multi-agent operation.
- GitHub Copilot CLI: terminal execution, planning, sessions, and remote operation.
- Local Agentic SDLC: parallel specialist work, session control, and integration ownership in one orchestrated local run.
- Copilot SDK: programmatic embedding and application-defined agent experiences.
- Copilot on GitHub: delegating issues to the cloud agent, with code review and Code Quality on every push.
- What's New in VS Code: current editor-side agent capabilities and workflows.
First-party sources for every claim are listed per slide in the speaker notes and in the references frontmatter above.