Framework-to-Business Application Workflow
Turn an abstract business framework into a practical decision map
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 6
- Confidence
- 96%
Begin with a clearly identified framework and ask a reasoning model to research its definitions, distinctions, and underlying mechanism. Give the model a concrete domain, such as marketing, manufacturing, or customer success, and resolve its questions about audience, geography, and time horizon. Next, require diagnostic criteria that make each category observable and comparable. Use those criteria to classify the domain's strategies, tactics, or problems, preferably in tables or scored charts. Finally, inspect ambiguous cases and decide what the classification means for hiring, automation, investment, or execution. The mechanism is concept-to-criteria-to-classification-to-decision: the framework supplies the lens, structured research supplies consistency, and human review preserves judgment where conditions are uncertain.
Origin
Extracted from Marketing Against the Grain while applying David Epstein's kind-versus-wicked problems framework from Range to HubSpot's marketing, sales, and customer-service domain.
Core principles
- 01Translate concepts into observable criteria before applying them
- 02Clarify the business domain and scope
- 03Classify activities consistently rather than relying on intuition
- 04Use structured outputs to expose assumptions and comparisons
- 05Preserve human judgment for ambiguous or changing environments
How to run it
- 1
Select the framework
Choose a business framework whose implications matter to a current decision. Identify its source and central distinction so the research stays grounded.
Pro tip Start with one framework rather than combining several conceptual lenses.
Watch out Do not apply a framework merely because it is popular.
- 2
Generate a deep-research prompt
Ask the reasoning model to write a prompt that deeply investigates the concept and applies it to your domain. Include the desired analyses, comparisons, tables, and charts.
Pro tip Have the model improve the research prompt before asking it to run the prompt.
Watch out Vague domain instructions can produce classifications that sound useful but lack operational relevance.
- 3
Define the operating context
Specify the market, audience, region, time horizon, and business function. Answer follow-up questions that materially affect the research scope.
Pro tip Treat the model's clarifying questions as a scope review rather than unnecessary friction.
Watch out Do not mix fundamentally different contexts in one classification without labeling them.
- 4
Build diagnostic criteria
Translate the framework's categories into observable criteria. For kind-versus-wicked problems, criteria can include rule stability, pattern repeatability, feedback speed, and the value of prior experience.
Pro tip Ask for criteria before asking for final classifications.
Watch out Category labels without explicit criteria encourage arbitrary judgments.
- 5
Classify the domain
Apply the criteria to the relevant strategies, tactics, roles, or problems. Use a table or scorecard to make comparisons and borderline cases visible.
Pro tip Request a rationale beside every classification or score.
Watch out Avoid treating a binary framework as proof that every case fits neatly into one bucket.
- 6
Turn the map into decisions
Review the output and identify implications for automation, human involvement, hiring, product design, or investment. Preserve human-machine collaboration where neither side performs best alone.
Pro tip Focus first on decisions that would change because of the classification.
Watch out Do not outsource the final strategic judgment to the model.
In the wild
The host supplied David Epstein's distinction between kind and wicked environments, then asked o3 to research and chart marketing, sales, and customer-success activities. The model created criteria such as rule clarity, pattern repeatability, feedback quality, and the effect of experience before sorting the work. The resulting map could inform where HubSpot should automate tasks and where AI should remain a copilot.
→ A reusable decision map connecting an abstract book concept to product and operating choices.
A manufacturer could define its domain as plant operations, supply planning, maintenance, and product development. The model could classify repeatable quality checks as relatively kind while treating disruption response or novel product strategy as more wicked, then identify different roles for automation and human judgment.
→ A prioritized automation plan that distinguishes stable processes from uncertain decisions.
Common mistakes
Skipping the diagnostic criteria
Jumping directly from a framework summary to recommendations hides how the model assigned categories. Define the decision criteria first so the output can be audited.
Leaving the domain underspecified
A classification can change with audience, geography, industry, or time horizon. Resolve those boundaries before running the analysis.
Treating the output as objective truth
A polished table is still an interpretation. Review assumptions and borderline cases before making consequential decisions.
Is it for you?
Best for
It is best for leaders and strategists who need to apply a book, theory, or management model to a specific organization quickly.
Not ideal for
It is not ideal when the source framework is poorly defined, unsupported, or irrelevant to the decision at hand.
From the transcript
“Then it was able to tell me and make a heuristic of how it would rate different things for kind versus wicked.”
“And then I was able to give it my domain, but you could give it to any domain.”
“imagine working with somebody at your company talking about a framework or a favorite book, and be able to come back with a treasure map…”
From the episode
ChatGPT o3 Deep Dive: 3 Wild New Use Cases