Query types you can try with Research Assistant

Please note: this feature is exclusive to ReadCube Enterprise.

Research Assistant works best when you tell it exactly what you want back, not only what you’re curious about. Below are eight kinds of queries researchers ask most often, each with a reusable blueprint and a worked example. Copy an example and swap in your own topic. Then adjust the output details until they match what you actually need.

A few blueprints reference @[your library]. Type @ in the query box and start typing your library’s name to point Research Assistant at your own papers instead of the published literature at large.

Check the evidence for a claim

When you need to know whether the literature backs a specific statement – your own hypothesis, or a claim from someone else – ask for the supporting and opposing papers side by side, along with the passage and caveats behind each one.

Blueprint: [Source] + [Claim] + [Per-paper output]

Example:

Find papers in @[your library] that support or challenge the following statement: “Liver stiffness above 8 kPa indicates significant fibrosis.” For each supporting paper, provide: the citation, the specific passage, and any caveats (population, study design, strength of evidence).

You don’t need a library to run this. Drop the @[your library] reference and Research Assistant will search the wider literature instead:

What does the published literature say about GLP-1 agonist use in adults with type 2 diabetes and chronic kidney disease, from the last five years? Cite the key sources.

Get background on a topic

Getting oriented on a topic outside your specialty doesn’t require reading widely first. Ask for the shape of it instead of the depth: what’s established, what’s shifting, and where people disagree.

Blueprint: [Topic] + [Population] + [Time window] + [Sub-topics]

Example:

Provide an overview of the published background literature on atrial fibrillation in adults over 65, from the last decade. Cover: epidemiology context, established understanding, and where the field is evolving. Cite the original research and flag any major sources of disagreement.

Aim the same query at your own library instead, and it gets shorter:

Based on @[your library], what do we know about beta-cell regeneration? Cite the supporting papers.

Compare two things

Use this pattern when you want the agreements and the differences between two things named explicitly, instead of buried across two separate summaries. Perhaps you are evaluating two drugs or two protocols, this prompt will help you compare items against each other using metrics you specify.

Blueprint: [A] vs [B] + [Focus aspects] + [Output format]

Example:

Compare pembrolizumab and nivolumab in the context of solid tumour immunotherapy. Focus on: mechanism of action, approved indications, and key safety differences. Return a short summary covering where they agree and where they differ.

Map a patent landscape

Skip reading patent filings one by one. This query surfaces who holds the intellectual property in a space and where the claims cluster.

Blueprint: [Technology] + [Time window] + [Output dimensions]

Example:

Analyze the patent landscape for solid-state battery electrolytes filed between 2020 and the present. Identify: the top 5 assignees by filing volume, the dominant claim focus areas, legal status (active, pending, expired), and geographic hotspots. Provide the citation for each patent referenced.

Find a method or protocol

Setting up a new experiment? Ask how others have run it, including the parameters and the pitfalls that don’t always get included in a published methods section.

Blueprint: [Method] + [Key parameters] + [Output format]

Example:

Find well-described protocols in the literature for measuring progeny survival in C. elegans. I care about: reagents, timing, and controls. Return: the step-by-step method as described, the key references, and any commonly noted pitfalls or failure modes.

Get a field overview

New to a field, or coming back to one after time away? This pattern gives you the shape of it – consensus, live debate, and open questions – all grounded with real research.

Blueprint: [Field] + [Named sub-topics]

Example:

Help me understand the current state of research on anaerobic digestion for biogas production. Cover: what is broadly accepted, where researchers disagree, which directions look most promising, and which questions remain open. Cite the key papers underpinning each point.

Build medical education content

This query connects two requests: turning a topic into a working outline, and turning a topic into a shortlist of speakers.

Blueprint (content outline): [Role] + [Source] + [Deliverable] + [Format specs]

Example:

You are a medical education content creator. Based on @[your library], create a content outline with references for a 45-minute CME presentation on the treatment landscape in chronic kidney disease, delivered by 2 expert faculty. Include: learning objectives, key sections with approximate time allocation, suggested figures or data points to highlight, and references for each section.

Blueprint (speaker suggestions): [Number] + [Programme specs] + [Per-expert output]

Example:

Suggest 5 potential expert speakers for a program with: topic: emerging therapies in endocrinology; target audience: adult and pediatric endocrinologists; format: 45-minute symposium. For each, include affiliation and 2-3 recent relevant publications.

Draft internal writing and admin

The internal writing that eats time without needing much originality – a briefing, a newsletter – and it doesn’t need to start from a blank page either.

Blueprint: [Source] + [Deliverable] + [Audience] + [Time window]

Example:

Based on @[your library], draft a 300-word internal briefing for our medical affairs colleagues covering papers from the last quarter. For each paper highlighted, include: title, one-sentence takeaway, why it matters for our audience, and the citation. End with a short paragraph on emerging themes worth watching.

A note on the brackets [ ]

Every example above is a recipe, meant to be used as a suggestion for your own prompts. The bracketed placeholders are yours to edit – narrow a time window, or spell out the exact output format you want. The more specific you are about what you want back, the closer the first answer will land to something you can use directly.