Seven prompts for research work with AI: structuring a question before you start, finding where to look, checking whether a source deserves your trust, distilling what you read, and pulling out the implications rather than a summary.
How to use these prompts
Copy a prompt into any capable AI model, then add your own material underneath it. Most of them are written to work on whatever you paste in: a plan, a draft, a decision you are weighing, a set of notes. The ones marked Assistant run as a conversation rather than a single exchange, so expect the model to ask you questions before it produces anything.
The prompts are deliberately short. They carry a method, not a script. Adapt the wording to your situation and the model you use.
Filter by what the prompt does
Showing all 7 prompts
Distillation
Summarizes a source into the format I need, preserving nuance.
- Synthesis
- Prompt
Summarize the following material into the form I need. Capture the core argument, key evidence, and conclusions, and drop the filler. Unless I specify otherwise, give me a short headline takeaway, then the key points, then anything notable or surprising. Preserve important nuance and flag anything you're uncertain about. The material is:
Insight Extractor
Pulls non-obvious insights and implications, not just a summary.
- Analysis
- Prompt
Go beyond summarizing the following material and pull out the non-obvious insights. Identify the implications, patterns, tensions, and 'so whats' that a careful reader would value but that aren't stated outright. For each, note what it means for me or my decision. The material is:
Landscape Map
Maps the players, positions, and white space in a field or market.
- Synthesis
- Prompt
Map the landscape of the following topic, field, or market. Lay out the main players, positions, schools of thought or segments, how they relate and differ, and where the tensions and white space lie. Conclude with what an informed observer should take away from the overall picture. The topic is:
Research Question Tree
Breaks a topic into a structured tree of answerable questions.
- Framing
- Prompt
Break the following research topic into a structured set of questions. Start from the central question, then branch into the sub-questions that must be answered to address it, and note for each what kind of evidence would answer it. Order them so the most decisive questions come first. The research topic is:
Source Credibility Check
Assesses reliability, bias, recency, and corroboration of a source.
- Evaluation
- Prompt
Assess the credibility of the following source or claim. Consider the author and their expertise, the evidence and method, recency, potential bias or motive, and whether it's corroborated elsewhere. Conclude with how much weight to place on it and what to verify before relying on it. The source or claim is:
Source Discovery
Identifies where to look and what’s worth reading for a question.
- Research
- Prompt
For the following research question, suggest where to look for high-quality information. Identify the types of sources most likely to hold the answer (reports, datasets, experts, publications, communities), name specific starting points where you can, and give search terms or angles for each. Flag where the best information is likely to be hard to find or paywalled. The research question is:
Pattern Detection in Qualitative Data
Six angles on a set of interviews or open responses: recurring language, emotional tone, contradictions, and the pairings that reveal a shared theme.
Contributed by Beth Kanter
- Analysis
- Prompt
Pretend you’re a detective. What patterns or clues can you find in this data about the interviewee’s preferences, challenges, or goals? What are the recurring words, phrases, or metaphors used by participants? Group them into thematic clusters Summarize each speaker’s emotional tone and underlying needs. Group similar emotional patterns. Identify any contradictions or tensions in what people say they want vs. what they do. Group by type of tension Cluster participant comments around behavior, motivation, and barriers. What themes emerge across interviews? Play matchmaker. Which quotes or behaviors from different interviewees would you pair together, and why? What shared themes connect the
Common questions
How can AI help with research without inventing sources?
Use it for the parts of research where invention is not a risk: structuring the question, identifying what kinds of sources would answer it, assessing the credibility of a source you already have, and synthesizing material you supply. The Source Credibility Check and Distillation prompts are built for that pattern.
What is the difference between summarizing and extracting insight?
A summary compresses what a source says. An insight names what follows from it that the source did not state. The Insight Extractor prompt asks specifically for the non-obvious implications, which is usually what you wanted when you asked for a summary.
How do I structure a research question for AI?
Break it into a tree of answerable sub-questions before you start searching. The Research Question Tree prompt does this, and the resulting structure tells you where the real uncertainty sits.