Chapter C15 · Interactive Course
Plan and decide with AI
Treat a generated plan as a draft to be made workable.
Core Learning Concept
Turn the proposed session into something people could actually carry out. Check the hard time limit, the order of dependent steps, transition time, and any unsupported assumptions. Then choose a small trial and a way to judge it. A plan is ready when its requirements work together, not when its heading sounds convincing.
What You’ll Learn in This Chapter
-
State the non-negotiable
Make a practical constraint explicit before asking for a plan.
A non-negotiable is a requirement the plan must satisfy, such as a thirty-minute room booking or no spending. State it before asking for ideas and check every proposed activity against it. A creative plan that violates the hard limit is not yet a usable plan.
-
Put steps in a workable order
Identify a dependency between actions.
A dependency is something a step needs before it can happen. People need the cards before writing on them; a room must be available before the session begins. Read the plan one step at a time and ask what must already be true. Move preparation earlier where necessary.
-
Check that the time fits
Verify a plan's total duration and transition time.
A duration label is a claim that needs arithmetic. Add the activity times, then include transitions such as handing out materials and regrouping. If a thirty-minute plan already contains thirty minutes of activities, it may have no room for movement or delays. Adjust the plan to fit the actual limit.
-
Compare on shared criteria
Use consistent criteria to inspect options.
Shared criteria let you compare options consistently. Use the same columns for cost, capacity, and access, and distinguish stated values from unknowns. A full-looking table may contain unsupported guesses. Leave missing evidence visible and investigate the gaps that could change the decision.
-
Name the assumptions
Separate supplied facts from assumptions in a recommendation.
An assumption is something treated as true without being established by the supplied evidence. Some assumptions matter more than others. If a plan depends on a room being available, confirm that before polishing the agenda. Label assumptions explicitly and prioritize checks by what would fail if they were wrong.
-
Try a small version first
Evaluate whether an AI-assisted process helps before expanding it.
A small trial tests whether the whole process helps in practice. Include preparation, generation, checking, and corrections when counting time. A fast first draft may require slow repair. Try one session or document, evaluate quality and total effort, and use that evidence before expanding the workflow.
-
Make a workable half-hour plan
Build a half-hour plan, fix its timing, identify an assumption and define a small trial.
-
Use AI for one everyday task
Prepare, inspect and improve an output for one real task, then state what the human contribution actually was.
In-Depth Practice & Real-World Application
Workplace Scenario: Using AI to draft project plans, timelines, and decision matrices as iterative working drafts.
Worked Example: Drafting Workable Schedules with Explicit Constraints
Common mistake: “Make a project plan to launch our new website.” (Produces an idealistic timeline with no dependencies or resource limits)
Effective pattern: “Draft a 6-week website launch plan for a 3-person team (1 designer, 1 developer, 1 marketer). Include weekly milestones, risks, and signoff gates.”
Framing resource constraints forces the AI to produce realistic, sequenced stages rather than fictional wishful thinking.
Key Takeaways
- AI project plans are starting baselines, not ready-to-execute schedules.
- Always ask the AI to list: “What could go wrong with this schedule, and what dependencies exist?”
- Use AI for risk assessment: “Act as a project devil’s advocate and challenge this rollout strategy.”
Workplace tip: After generating a plan, prompt: “Identify the 3 weakest assumptions in this timeline and suggest contingency buffers.”