What you will learn
- Summarise a long document without losing its main argument
- Compare two or more sources on the same question
- Turn scattered notes into a structured research brief
- Extract quotes, facts, and claims for later checking
- Spot gaps and disagreements between sources
- Organise findings so you can cite them later
Begin with the question, not the pile of documents
Research goes wrong when it starts with collecting everything and ends with a pile nobody reads. Write the research question down first, in one sentence, and list the three or four sub-questions that would answer it. That list becomes your filter: a document is useful if it answers one of those sub-questions, and skippable if it does not.
Share that question with Claude at the start. Ask it to organise the work around your sub-questions rather than producing a general overview. You will get a tighter result, and you will notice faster when a sub-question has no good answer yet.
Summarise one document at a time
For a single document, ask for a summary that keeps the structure: the main claim, the evidence offered for it, the limits the author admits, and anything the author leaves unresolved. A summary that only lists topics is less useful than one that preserves the argument. If the document is long, summarise it in sections and then ask for a short overall summary of those section summaries.
Ask for the summary at the length you actually need. A two-paragraph brief suits a meeting note; a page of structured notes suits a report you will write from. State the audience too, because a summary for a specialist and a summary for a newcomer emphasise different things.
Compare sources instead of stacking them
The real value of research is comparison. Give Claude two or more sources and ask where they agree, where they disagree, and what evidence each side offers. Ask which claims appear in only one source, and which claims all sources share. That comparison is often where the interesting question for your own work appears.
Be careful with agreement that is only apparent. Two sources may use the same words to mean different things, or cite the same original study without adding new evidence. Ask Claude to note when sources share an underlying source, so a single finding does not look like a consensus.
Summarise this article in about 300 words, keeping its structure: main claim, evidence, limits, and open questions. Then list five factual claims from it that I should verify against the original source before quoting them, and tell me which parts are the author's opinion rather than evidence.
Build a research brief you can reuse
Ask Claude to turn your summaries into a brief with a fixed shape: the question, the answer in brief, the evidence by sub-question, the disagreements, and the gaps. A fixed shape makes later research comparable, because each new brief answers the same headings. Keep the brief short enough to read in one sitting.
Attach your own notes to the brief: which sources you trust most and why, which claims you have verified, and which questions remain open. Claude's output is a draft; your judgement about source quality is what turns it into something you can stand behind.
Verify before you quote
Summaries are a starting point, not a source. Before you quote a fact, a statistic, or a claim in your own work, check it against the original document. This matters most for numbers, dates, names, and anything that sounds surprising. If Claude summarised a document you pasted, the original is right there; use it.
Also check the date and context of each source. A finding from five years ago may still hold, or it may have been overtaken. Note the publication date in your brief so future readers, including you, can judge freshness at a glance.
Tip: Ask Claude to separate fact from opinion in every summary. A brief that mixes the author's claims with the author's evidence will mislead you later, when you no longer remember which was which.
Projects for longer research
For research that runs over days or weeks, a Claude project keeps your documents and instructions in one place, so each new question builds on the same material instead of starting from scratch. Upload the core sources, write a short note on what the project is about, and add new documents as you find them.
Review the project material regularly. Remove sources that turned out to be weak, and add a note when your question changes. A tidy project stays useful; an untended one quietly accumulates material you no longer trust.
Frequently asked questions
Can Claude summarise a long PDF or article?
Yes. Share the document or paste its text and ask for a structured summary that keeps the main claim, the evidence, and the limits. For very long documents, summarise section by section.
Can I trust Claude's summary without reading the original?
For getting oriented, yes. For quoting facts, statistics, or claims in your own work, check them against the original document first.
How do I compare several sources with Claude?
Give it the sources together and ask where they agree, where they disagree, and what evidence each side offers. Ask it to flag claims that appear in only one source.
What is a research brief, and why make one?
A short document with a fixed shape: the question, the brief answer, the evidence, the disagreements, and the gaps. It makes findings easy to reuse and easy to check later.
Does Claude know about recent events?
Its knowledge has limits, and details change over time. For current facts, paste the source material yourself rather than relying on what it recalls, and check dates carefully.