Deep research
Definition
Deep research is a mode in AI assistants where the model runs many web searches on its own, reads the sources it finds and writes a longer report with citations. It takes longer than a normal answer and suits questions that need several sources compared.
By Ronan Liedmeier, fractional head of growth and course author · Published 17 October 2026 · Last updated 17 October 2026
Also called: Research mode, ChatGPT deep research
ChatGPT calls it deep research and Claude calls it Research. It fits decisions with real stakes, such as choosing a supplier, where a person would otherwise spend an afternoon reading.
How it works
You describe what you need. ChatGPT may ask clarifying questions and draft a research plan you can edit before it starts, and OpenAI describes the result as a structured report with citations or source links so you can verify it. Claude's help center describes Research as running multiple searches that build on each other, with Claude deciding what to look into next based on what it finds.
Give it a clear brief: state the decision you are making, list numbered questions, set the time range for sources and name the output format.
Example
An ecommerce manager is comparing two parcel carriers for shipments to Germany. Her brief lists her order volume, five numbered questions on rates, delivery times and claims handling, and asks for a comparison table using sources from the last 12 months. The report favors one carrier on speed. When she opens the citation behind the delivery figure, it turns out to describe that carrier's service in France.
Common mistake
Trusting the citations without opening them. A report can cite a real page that does not say what the report claims, or rely on prices and features that are years out of date. Open the sources behind any claim your decision rests on, check their dates, and have a qualified professional confirm anything on law or tax.
Learn it in AI Academy
- In-Depth Research With AIBeginner · 12 min
Related terms
- Hallucination
- A hallucination is a false statement an AI model presents as fact, such as a made-up statistic, quote or source. It happens because language models are trained to produce likely-sounding text, and likely-sounding text is sometimes wrong.
- AI agent
- An AI agent is an AI system that works toward a goal by choosing its own next steps: it picks a tool, uses it, looks at the result and repeats until the task is done. You set the goal and the limits, and the agent works out the path.
- Knowledge cutoff
- A knowledge cutoff is the date after which an AI model has no information from its training. Anything that happened later, such as news, price changes or new product releases, is unknown to the model unless it searches the web or reads a document you give it.