Skip to content
SES-L2

The Principles of AI Collaboration: Context, Planning and Review

Understand how to provide direction and context, plan in proportion to the task, review results, and preserve useful knowledge and methods. Explanations, short demonstrations and discussion.

See the course

Upcoming dates

No dates are scheduled for this course right now. Tell us and we will notify you when a date opens.

What you learn

Can explain which information matters, when a plan helps, how to identify an unsupported claim, what done means, and when a method deserves a skill.

What you need first

Familiarity with Session Level One, “Beyond Chat: Understanding and Working with AI,” or equivalent knowledge. Workshop Level One is not required.

Curriculum

10 chapters

  1. 1

    Give Your Work a Direction

    Pause to name your goal, known facts and what a good result looks like before asking AI, and treat AI as an adviser rather than only an instruction-follower.

  2. 2

    Give AI Useful Context

    Compare a vague request with an informed one, sort known facts from gaps, and watch for examples that quietly leak facts you don't want reused.

  3. 3

    Prepare Your Session

    Select relevant saved files explicitly and compare the answer with what was actually read.

  4. 4

    Know When to Plan

    Distinguish a small direct task from multi-step work that needs a plan first, and review that plan before letting AI execute it.

  5. 5

    Build a Useful Plan

    Start a plan from the expected outcome and confirmed facts, break it into a few small checkable phases, and name limits early.

  6. 6

    Review Before You Begin

    Check a plan from three viewpoints, bring in a second independent reader, and correct unsupported claims with a specific, justified fix.

  7. 7

    Save the Decisions That Matter

    Save approved decisions and progress as they emerge. Record sources and check dates, then update stale facts without losing useful history.

  8. 8

    Execute and Continue with Clear Context

    Decide when a fresh session helps versus continuing the same one, load the approved plan before executing one phase at a time, and check a changed fact against each step rather than rewriting everything.

  9. 9

    What Does “Done” Really Mean?

    Distinguish produced, checked and delivered, agree success criteria before starting, and ask what's complete, missing and actually sent rather than trusting a confident summary.

  10. 10

    Know When to Create a Skill

    Decide when a recurring method deserves a skill. Separate fresh task facts from reusable steps and intentional company or project rules.