One of the most useful things we can do with AI is not simply ask it for answers, but use it to examine ideas, hypotheses, business models, decisions, and problems through different intellectual frameworks.
I've become increasingly interested in this approach to thinking.
When I come across an insightful video, article, podcast, or book, what interests me most is not necessarily the conclusion. It's the structure behind it.
How does this person think? What steps do they follow? Which assumptions do they test? What perspectives do they use to examine a problem? What criteria guide their judgment?
In other words, I'm interested in the architecture of their reasoning.
Because when that architecture is strong, it becomes something we can use.
From consuming ideas to extracting frameworks
We spend a remarkable amount of time consuming other people's thinking.
We watch an insightful interview, read a compelling analysis, or listen to someone break down a complex problem. We recognize that there's something valuable there. Perhaps we save the link, take a few notes, or share it with someone.
And then, more often than not, we move on.
But what if we treated valuable intellectual content differently?
Instead of preserving only its conclusions, we could extract the framework that produced them.
We could identify the sequence of questions, the evaluation criteria, the underlying assumptions, and the different perspectives involved. Then document that structure in a reusable format — even something as simple as a Markdown file.
The next time we face a similar problem, we wouldn't need to reconstruct the original insight from memory.
We could apply the framework directly.
Consider business idea validation.
Rather than asking AI whether a business idea is good, we could take that same idea and examine it through several independent frameworks.
A validation framework might reveal weak assumptions about customers and demand.
A strategic framework could expose competitive disadvantages or overlooked opportunities.
An architectural framework might show that the solution itself is poorly structured.
A risk model could uncover vulnerabilities we hadn't considered.
A scalability framework might identify what would break if the idea actually succeeded.
The important part is that these frameworks do not necessarily agree.
And that's precisely what makes them useful.
Each one reveals a different dimension of the problem. Together, they create a richer picture than a single model-generated opinion.
From individual prompts to thinking systems
This becomes even more interesting when the process is automated.
Imagine describing a business idea in a voice message.
A system transcribes it, extracts the relevant information, and organizes it into a structured format.
That information is then passed through several frameworks, either sequentially or in parallel.
Each framework examines a different aspect of the idea. The results are collected, compared, and organized into a coherent analysis.
Weak assumptions become visible. Risks are identified. Opportunities emerge. Contradictions between perspectives can be examined rather than ignored.
Instead of receiving one answer from AI, we receive a structured examination of the idea through several distinct systems of reasoning.
And the same approach extends far beyond business validation.
It can be applied to strategic decisions, product development, organizational challenges, personal projects, or almost any recurring problem that benefits from being examined from multiple perspectives.
What matters is not the complexity of the automation.
What matters is the shift in how we understand AI's role.
We stop treating it as an authority that produces answers and start treating it as a system that helps us work with different ways of thinking.
This does not remove the need for judgment. Frameworks have their own assumptions and limitations, and their outputs still need to be questioned.
But they can make our reasoning more explicit, more systematic, and easier to examine.
The real opportunity
I think this is one of the most underestimated ways to use AI.
Not as a replacement for thinking.
Not merely as a text generator.
And not as a machine for producing quick answers.
But as infrastructure for thinking.
The future value of AI may depend not only on how much more capable models become, but also on how effectively we learn to build systems around them.
Systems that preserve useful intellectual frameworks, connect different perspectives, and help us examine our own assumptions.
The opportunity is not simply to make AI think better for us.
It's to build better conditions for thinking ourselves.
Filed underHuman Systems
