Context-First AI Conversation
Make AI interview you before it recommends an answer
- Difficulty
- Starter
- Time to result
- ~days to results
- Steps
- 5
- Confidence
- 99%
Instead of treating an AI assistant like a search box, treat it like a capable colleague who needs to understand the situation before advising you. Begin with the outcome or question, then explicitly instruct the AI to ask one question at a time until it has enough context. Answer those questions with the constraints, history, goals, and trade-offs a human adviser would need. Only then ask for a recommendation. If the first answer misses the mark, explain what context it overlooked and continue the exchange rather than discarding the tool. The mechanism is simple: a guided interview converts hidden knowledge in your head into usable context, which gives the model better inputs and produces a more relevant output.
Origin
Joshua Wöhle teaches this as a foundational Mindstone technique for replacing search-engine-style prompting with person-like AI conversations.
Core principles
- 01Treat AI interaction like a conversation with a capable person
- 02Answer quality cannot exceed the quality of the supplied context
- 03Questions expose missing information before advice hardens
- 04Feedback should continue the conversation rather than end it
How to run it
- 1
Define the outcome
Tell the AI what decision, deliverable, or problem you ultimately want help with. Do not ask it to solve the problem yet.
Pro tip Include what a useful final answer must enable you to do.
- 2
Request an interview
Ask the AI to gather the context it needs by asking one question at a time before answering.
Pro tip One-at-a-time questions make gaps easier to notice and answers easier to refine.
Watch out Do not let the AI jump straight to a recommendation.
- 3
Supply concrete context
Answer with relevant goals, constraints, prior attempts, stakeholders, and evidence. Treat the exchange as you would a briefing with a skilled colleague.
Watch out Vague answers recreate the same context problem.
- 4
Generate the answer
When the interview is complete, ask the AI to synthesize the context and produce its recommendation or work product.
Pro tip Ask it to connect its recommendation to the context you supplied.
- 5
Continue with feedback
If the result is off target, identify the missing or misunderstood context and ask for a revision. Keep the conversation alive until the output fits the situation.
Watch out Do not interpret one weak answer as proof that AI cannot help.
In the wild
A founder asks the AI to compare two growth strategies, but first has it interview them about runway, customers, team capacity, margins, and risk tolerance. The AI then recommends one strategy and explains how the supplied constraints shaped the choice.
→ The recommendation becomes specific to the company instead of generic business advice.
Common mistakes
Using AI like a search engine
A short query omits the context that any competent human adviser would need, producing a superficial answer.
Abandoning the first weak answer
Rejecting the tool instead of correcting its missing context prevents the conversational improvement loop.
Is it for you?
Best for
It is best for decisions, strategy, advice, and other tasks whose answer depends heavily on context.
Not ideal for
It is unnecessary for simple factual lookups or tightly specified mechanical tasks.
From the transcript
“one is I am I am a technologist by by nature, I'd say.”
“you'd ask me 10 questions first to get that context.”
“You cannot get a perfect answer if you do not give the context.”
From the episode
Joshua Wöhle: Turn AI Into Your Competitive Advantage and 50X Your Productivity
Joshua Wöhle