Most people use AI to look things up. That is the lowest-value use of the technology. The highest-value use is as a thinking partner — a system that challenges your assumptions, argues against your plan, and surfaces what you have not considered. That shift in how you prompt changes what you get back.
The search engine trap
Using AI to answer factual questions is marginally better than Google. The model hallucinates. The information may be outdated. And you are not using the model's actual strength, which is reasoning over your specific context.
When you ask "what is the best pricing model for a SaaS product," you get a generic answer that applies to everyone and therefore helps no one. The model has no idea what your churn looks like, who your buyers are, or what your competitors charge. It is pattern-matching on training data, not thinking about your situation.
The search engine trap is comfortable because it feels productive. You get an answer quickly. But the answer is rarely the one you needed.
What a thinking partner does
A thinking partner challenges your assumptions. It argues against your plan. It identifies your blind spots and asks the question you did not think to ask.
AI can do all of this — if you prompt it correctly. The difference is not the model. It is the framing. Most prompts are requests for information or output. Thinking partner prompts are requests for critique, stress-testing, and adversarial reasoning.
The model is very good at this. It has seen thousands of plans succeed and fail. It knows the common failure modes. It will tell you, if you ask.
The steelman prompt
The single most useful prompt for a solopreneur:
Here is my plan: [plan]. What is the strongest argument against it?
This forces the model to argue against you. Not a weak objection — the strongest possible case against your plan. That is the steelman: the most charitable, most rigorous version of the opposing argument.
Why this works: you are too close to your own plan to see its weaknesses clearly. You have been thinking about it for weeks. The model has no attachment to it. It will find the load-bearing assumption you glossed over, the market condition you assumed away, the competitor move you did not account for.
Run this before you commit resources. Run it before you pitch a client. Run it before you publish.
Pre-mortem prompting
The pre-mortem is a technique from decision research. Instead of asking what could go wrong, you assume it already did.
Assume this project failed completely. It is six months from now and the project is dead. What went wrong?
This prompt works because it bypasses optimism bias. When you ask "what could go wrong," your brain generates a polite list of manageable risks. When you assume failure has already happened, your brain works backward from a concrete outcome and finds the real causes.
In practice, the pre-mortem surfaces things the steelman misses: execution failures, team dynamics, timing problems, customer behavior that did not match the assumption. It is consistently more useful than asking what could go right.
Run it on any project before you start. The output is not a reason to stop — it is a checklist of what to watch.
The assumption audit
Every plan rests on assumptions. Most plans fail because the assumptions were wrong, not because the execution was poor.
Here are the assumptions my plan depends on: [list]. Which of these is most likely to be wrong, and why?
This prompt forces the model to evaluate your assumptions rather than accept them. Feed it a real list — not the obvious ones, but the ones you are quietly confident about. Those are the dangerous ones.
A concrete example: a content strategy that assumes organic search traffic will compound over 12 months. The model will flag that this assumption depends on algorithm stability, competitive response, and indexing behavior — none of which you control. That is useful. It does not kill the strategy; it tells you where to build contingencies.
The assumption audit is most valuable when you are about to make an irreversible decision.
How to use AI for strategy sessions
Before a client strategy session, run a 20-minute AI thinking session. Feed the model the client context, your proposed approach, and ask it to steelman the alternative approaches you are not recommending.
The prompt:
I am recommending [approach A] to a client. The alternatives I considered and rejected are [approach B] and [approach C]. What is the strongest case for each alternative, and what would have to be true for one of them to be the better choice?
You arrive at the session with your thinking already stress-tested. You know the objections before the client raises them. You have already decided whether they change your recommendation or not.
This is not about hedging. It is about arriving prepared. A client who asks "did you consider X" gets a direct answer, not a pause.
We send our thinking partner prompt templates — steelman, pre-mortem, assumption audit — to Field Notes subscribers. Get them at avakata.agency/contact.html.
The doing is the easy part
AI is very good at doing. Writing, summarizing, formatting, scheduling, drafting. The doing is the easy part — the model handles it reliably and fast.
The thinking is harder. What to do, why, and whether the plan is sound. That is where a solopreneur has the biggest gap: no team to push back, no colleague to say "have you considered," no second opinion before the decision is made.
AI fills that gap if you use it correctly. Not as a search engine. Not as a writing assistant. As the thinking partner you do not have on payroll.
If you want to work through how this applies to your specific situation, book a discovery call. We will run a thinking session on whatever decision you are sitting on.