Deep Research is a powerful new agentic AI tool on Scopus AI. It draws on our proprietary reasoning engine to break down your query and then independently set objectives, search and iterate as it uses the rich, curated literature in Scopus to explore each component. Findings are synthesized and shared in a multi-page report. This nuanced analysis saves days of preliminary research and surfaces connections and perspectives that can help accelerate your next discovery.
Agentic AI is still a very new technology, so we’ve asked the Deep Research development team for some tips to help you get started.
Although Deep Research is happy tackling many types of queries, its advanced capabilities make it a great companion for the challenging questions other AIs struggle to answer. Deep Research has been instructed to break your question down into simple components and identify which search option(s) will get the most relevant results – it can opt for vector search, keyword search, or both. It then uses these to scan the curated abstracts in Scopus.
*As Deep Research searches, it adapts its approach, based on the new information it surfaces. Findings are delivered in the form of a nuanced analysis in the Deep Research report.
The great thing about Deep Research is that you don’t need to worry about keywords or Boolean search strings—just ask your question in natural language. To get the best results, include enough context to help Deep Research interpret your goal. You can also combine multiple directions or questions in a single prompt to guide the exploration.
Here’s an example of a well-constructed question with a clear objective:
“Which existing anti-inflammatory drugs could be repurposed for slowing the progression of Parkinson’s disease, and what experimental protocols would validate this approach?”
Want to tailor the scope of your search? No problem. You can instruct Deep Research to search by:
All you have to do is indicate your chosen specifications when you type in your natural language query.
Here’s an example:
Identify novel epigenetic targets for liver fibrosis based on current preclinical evidence and focus on studies from 2020-2024
Want to learn more about one of the findings in the report? Or maybe it’s sparked a new line of thinking you’d like to explore? Our Conversational follow-up feature enables you to ask additional questions. Deep Research analyzes whether it has enough information to answer your question directly, based on the research it’s already completed. It may also run a new search to find exactly what you need.
After a report is generated, you can ask it to elaborate on specific points and choose whether to use Deep Research again or switch to a standard Scopus AI search. While you can’t edit the report itself, asking for changes will trigger a new report. Deep Research also uses what it’s learned during its research phase to decide what to explore further in subsequent steps.
Deep Research reports are designed to inspire critical thinking and creativity. Given user feedback and in alignment with our principles, the tool is carefully designed to avoid mimicking important areas that require human creativity and input, for example a traditional literature review. That’s because we don’t want to replace the work of researchers, we want to support it, with insights that will guide you toward your next discovery.
A Deep Research report includes:
Together, these provide ideas and suggestions for further exploration, and every claim is referenced – this gives you the reassurance that only trusted, peer-reviewed insights from Scopus have been used. It also enables you to jump to the source material and continue your research journey.
The main body of a Deep Research report breaks down the knowledge the AI tool has found into 10-20 individual components. And instead of making factual statements, the report shows you information that will help you explore and understand these components. This includes a synthesis paragraph, which suggests actionable ways you can use and connect these areas to further your research.
So, we’ve designed Deep Research to not only answer your query, but to deliver a much broader view of the research landscape and some unconventional ways of looking at it – hopefully generating some serendipity in the process.
*Deep Research currently uses the abstracts of documents in Scopus published since 2003. We are already busy looking at options to expand the sources it can draw on.
