Ask Cheiron is at the core of your Cheiron experience. Pose a question or a task — anything from a quick literature lookup to a multi-step competitive analysis — and Cheiron does the research and synthesis you'd otherwise farm out to a strong analyst. Behind every answer is a real-time scan across the corpora your team trusts: PubMed, ClinicalTrials.gov, FDA, EMA, ICH, MFDS, USPTO, curated peer-reviewed literature, and more. For complex queries, Cheiron may review over thousands of documents to find the strongest evidence.
What you get back isn't a list of links. It's a structured answer fitting to your task, with claims, evidence, and citations. Every claim is verifiable down to the passage it came from: see our Deep Citations guide for how that works.
How to ask Cheiron
1. Write a clear question
Cheiron understands plain English. The more specific your question, the more focused the answer. A good question usually includes:
Objective: what you're trying to learn or produce
Scope: molecule, indication, mechanism, geography
Constraints: phase of trial, year range, regulatory body, evidence type
Examples by function:
Clinical: "What primary endpoints have been used in recent Phase 3 trials for major depressive disorder?"
Regulatory: "What FDA precedents exist for accelerated approval based on PFS in NSCLC since 2018?"
Medical Affairs: "Summarize the most recent Phase 3 efficacy and safety data for GLP-1 agonists in NASH, and contrast against semaglutide."
R&D: "What are the current resistance mechanisms reported for KRAS G12C inhibitors?"
Commercial: "Compare 2024 launch performance across the BCMA-directed therapies in multiple myeloma."
If you don't know exactly what to ask, the sample queries below the input box are tailored to your function and a good place to start.
2. Let Cheiron run
After you submit, Cheiron runs role-aware pipelines across the relevant databases. Depending on the question, this can mean scanning thousands of documents to find the strongest evidence. The progress messages show you what Cheiron is doing at each step.
You can click on sources directly here too to get a preview of ....
3. Read the structured answer
Once the search is complete, Cheiron presents a structured answer based on the most relevant evidence. Treat it the way you'd read the work output of a strong colleague: skim the structure, drill into the parts that matter for your decision, verify before you act.
The references panel on the right lists every source the answer drew from. Click a sentence in the answer to filter the panel to just the sources behind that claim, or click any source card to read the cited passage in the original document. See the Deep Citations guide for the full verification workflow.
4. Iterate
If the answer doesn't quite land, ask a follow-up rather than starting over. Cheiron treats each conversation as a thread, so follow-ups carry the prior context — you can refine ("focus on Phase 3 only"), expand ("now compare against EMA precedents"), or pivot ("what about the safety side?") without restating the original question.
For deeper iteration on a specific source, open Document Chat from any source card.
Tips for getting better answers
Be specific about scope. "What's the evidence for X in Y" is much stronger than "tell me about X." Cheiron does better with constraints than with breadth.
Start with the question, not the action. "What are the resistance mechanisms for KRAS G12C inhibitors?" works better than "Find papers about KRAS G12C." Cheiron is closer to a research analyst than a database search engine.
Use real domain language. Pharma vocabulary helps Cheiron home in. "Accelerated approval based on PFS" is more precise than "fast-track FDA approvals based on tumor shrinkage."
Iterate, don't restart. If your first answer didn't hit the mark, ask a follow-up rather than starting a new conversation. Cheiron remembers context.
Verify before you act. Every claim is linked to its source for a reason. The thirty seconds you spend in Deep Citations is what makes Cheiron's answer something you can stake your name on.