Who this track is for: You use AI regularly to get things done — writing prompts, building with assistance, working through problems step by step. This track moves you from doing work with AI toward defining the work clearly and handing entire tasks off — so the bottleneck isn't you anymore.
Level 1
Dabbler
Starting Point
Doer
Goal
Delegator
Advanced
Director
Expert
Disruptor
Session 1
The Delegator's Toolkit
Shared vocabulary and foundational delegation frameworks
How AI Behaves
- Three behaviors that explain most AI surprises
- Why the same prompt produces different results — and how to account for it
The Three Pillars of a Clear Request
- Scope — what you're asking AI to do
- Intent — why it matters, what success looks like
- Structure — the format you expect back
Delegation Contracts
- User stories with Given/When/Then acceptance criteria
- Defining "done" before you start, not after
Standing Instructions
- Project context files that persist across conversations
- Eliminating the "re-explain everything" problem
Session 2
Define the Work
Structure large goals into delegatable pieces with consistent execution
Decomposition
- Breaking big goals into independently shippable story-sized pieces
- Managing a backlog AI can act on, one story at a time
Skills: Encoding Your Judgment
- Capturing repeatable processes as reusable AI instructions
- The "We Do, You Do" pattern: build a skill from something you just practiced
- Why skills solve the "different results every time" problem
Manual Review as a Quality Gate
- Checking AI output against acceptance criteria — pass or fail
- No vague "looks good" calls: explicit pass/fail against the spec
Session 3
Verify & Ship
Automate your quality gate and deploy to a live URL
Acceptance Criteria as Test Specs
- Given/When/Then criteria map directly to automated test structure
- Having AI generate tests from the spec you already wrote
The Closed Loop
- Criteria → tests → fail → implement → pass
- Why manual re-checking doesn't scale (the two-week cliff)
- Visual verification: checking what users see, not just what code does
Shipping
- Tests gate deployment: verify it, then it ships
- Deploying a tested application to a live URL
Session 4
Scale Your Delegation
Run parallel workstreams and stop being the bottleneck
Delegation Judgment
- Assessing which stories are ready to delegate vs. need more thought
- The delegation-ready test: is "done" defined? Is context loaded?
Background Execution & Sub-Agents
- Running tasks in the background while you work on something else
- How sub-agents handle specialized work within a single task
- Why each task gets a fresh conversation — and why that matters
Trusting Your System
- Batching similar stories and delegating in parallel
- Letting automated tests verify results instead of watching every step
By the End of This Track, You Can…
✓
Write user stories with acceptance criteria that define "done" before a single line of code is written
✓
Decompose a large feature into independently shippable, delegation-ready stories
✓
Encode your team's repeatable processes as skills — so AI follows them consistently every time
✓
Have AI generate automated tests directly from your acceptance criteria
✓
Deploy a live, tested application — gated by passing tests, not manual sign-off
✓
Delegate multiple tasks in parallel and trust your system — not your attention — to catch problems