Research and Thinking
Use Claude as a thinking partner, not a search engine. Progressive research that builds compounding knowledge.
Most people use Claude like a search engine with better grammar. Ask a question, get an answer, close the tab. That leaves most of the value on the table.
Ask "What's the market size for legal tech?" and you get a number. Ask "I'm considering entering legal tech with an AI intake product for personal injury firms. What are the three biggest risks I'm not seeing, and which of my assumptions might be wrong?" and you get a thinking conversation. You gave Claude your context and asked it to challenge you.
The mindset shift: Claude is not answering your question. It is thinking with you about your problem. You bring the context. Claude brings the synthesis, the challenges, and the angles you haven't considered.
The Progressive Research Session
The most powerful research technique: a conversation where each turn builds on the last.
Turn 1, start broad: "What are the main approaches to [topic]? Give me a structured overview with the key trade-offs."
Turn 2, drill in: "Expand on [the second approach]. What are the real-world implementation challenges? What separates solutions that work from ones that don't?"
Turn 3, apply to your situation: "Given all that, I'm [team size, timeline, constraints]. Where's the gap in the market?"
Turn 4, challenge the synthesis: "What's the strongest argument against the conclusion you just reached?"
By Turn 4 you have an analysis built for your situation that would have taken days of independent research. You're not asking four questions; you're running one progressive analysis.
Research Prompts for Common Tasks
Competitive analysis:
"I'm building [product] for [audience]. Who are the top 5 competitors? For each: what they do well, where they fall short, what gap they've left open. Which gap is most worth targeting, and why?"
Decision research:
"I'm deciding between [A] and [B] for [context]. My constraints: [list]. Walk me through the key trade-offs, what I'd need to believe for each option to be the right call, and which assumptions in my framing might be wrong."
Unknown territory:
"I know little about [topic]. Give me the mental model I'd use to think about this domain correctly, not a Wikipedia summary. Then the three things experts argue about most, and why."
The Key Habit: Argue With Claude
The most underused Claude skill is productive disagreement. Most people accept the first answer. Push back instead: "I disagree with your second point because X." Claude will engage and either update its position or explain why it holds. Either way, you learn something.
Correct its assumptions: "You're assuming X, but in our market Y is true. Recalculate from there." Or stress-test: "That's too optimistic. Give me the pessimistic scenario."
The most powerful form is steelmanning:
"I'm leaning toward [your position]. Steelman the opposing view. Give me the strongest possible argument for [the alternative], including the evidence I'm probably underweighting."
Stop using Claude to confirm what you believe. Use it to challenge what you believe.
Progressive research builds compounding knowledge: start broad, drill into specifics, apply to your context, then challenge the synthesis. Use structured prompt templates for competitive analysis, decision research, and learning new domains. And argue with Claude: push back, steelman the opposing view, ask for the pessimistic scenario. The best insights come from the back-and-forth.