Frameworks, the inside of a smart plan

Two founders run the same interview with the same PM agent. One gets a pipeline with four stages named after nothing. The other gets AIDA, with a conversion metric per stage and a dashboard that shows where leads die. Same model, same prompt. The difference is what was inside the plan.

The universities already wrote down the best thinking on revenue, talent, operations and innovation, and the model has read all of it. Prompt engineering was people describing, in their own words, things Harvard and Stanford already described properly. You do not need to memorize a single framework. You need to know they exist, and ask AI to teach you the one you need, when you need it.

Smart plan, average slop

Two founders run the same interview with the same PM agent. One gets a pipeline with four stages named after nothing, and a dashboard of counts. The other gets AIDA, with a conversion metric per stage, and a dashboard that shows exactly where leads die. Same model. Same prompt. The difference is what was inside the plan.

This is the thing nobody tells you about AI output: the model averages over everything it has read unless you point it somewhere specific. Point it at a framework and it stops averaging and starts applying.

You do not memorize them, you know they exist

The universities already wrote down the best thinking on how businesses grow, hire, operate and invent, and the model has read all of it. Your job is not to learn a hundred frameworks. Your job is to know they exist, so that when the interview asks what stages a lead moves through, you reach for AIDA instead of making something up.

• Talent
  How you interview, onboard, develop, grow and exit people. Situational leadership, the nine-box grid, the employee lifecycle, psychological safety.
• Revenue
  How you find, convert and keep customers. AIDA, STP, Jobs to Be Done, the Empathy Map, the Business Model Canvas, Porter's Five Forces.
• Operations
  How the work gets done and measured. OKRs, SCQA, the Pre-Mortem, value stream mapping, the theory of constraints.
• Innovation
  How you find the next thing. Design Thinking, the Lean Canvas, Blue Ocean, the innovator's dilemma, discovery-driven planning.

The master prompt pattern

One sentence, reused forever: I need to [task]. Teach me [Framework] from [University]. Ask it in the context of your business, not in the abstract. Read what comes back. Write down the one sentence you will use. That sentence goes into the interview, and from there into the plan, and from there into the build.

Then the plan names the framework. When docs/plans/build-1-plan.md says the pipeline follows AIDA, every agent that reads it builds a pipeline that follows AIDA, and the QA agent tests it against AIDA. The framework is not a slide you saw once. It is the vocabulary of the system.

The frameworks your CRM already is

You built these today. Name them and the plan gets sharper.

• The pipeline is AIDA or STP
  Attention, interest, desire, action. Or segment, target, position. Either way the stages have names that mean something, and a conversion metric between each.
• The interview is Jobs to Be Done
  What job is the customer hiring you for? Ask it that way and the inquiry types, the custom attributes and the nurture emails all write themselves.
• The dashboard is OKRs
  Six numbers you would act on Monday are six key results. Name the objective above them and the dashboard becomes a plan, not a report.
• The Definition of Done is a pre-mortem
  Imagine it shipped and failed. Why? Every answer becomes a line in the Definition of Done before the build starts.

One framework taught to you by AI in the context of your own business, one sentence carried into the plan, and a plan that names its framework so every agent builds to it. You have felt the difference between a smart plan and average slop, and it took ten minutes.