Every physician in Canada knows the CPS. The Compendium of Pharmaceutical Specialties has been the definitive drug reference in this country for decades. It’s the big blue book that sits on the shelf in every clinic, every pharmacy, every residency program. If you trained in Canada, you learned to trust it.
Drug Mode is built on it — and on the other references Canadian physicians rely on every day.
When we started building Asklepius, the pharmacotherapy question was obvious and urgent. Physicians were already using ChatGPT for drug questions, and the answers were often wrong in ways that were hard to catch. American dosing. Hallucinated interactions. No sources. Confident tone regardless of accuracy. We needed to build something grounded in the references Canadian physicians already trust, and make it work the way they actually need it to at the point of care.
Why six tools instead of one
The most common drug questions in clinical practice aren’t general. They’re specific workflows. You’re not usually asking “tell me about metformin.” You’re asking “can I combine these two drugs safely,” or “how do I switch this patient from one SSRI to another,” or “what should I consider deprescribing in this 84-year-old on eleven medications.”
These are fundamentally different questions. They need different data, different structure, different sources. A single search box forces the model to guess what you’re actually after. So we built six separate tools, each designed for a specific clinical workflow.
Info gives you a structured drug monograph. Dosing, indications, contraindications, monitoring parameters. The information you’d find if you opened the blue book, organized for how you actually use it at the point of decision.
Interactions checks drug-drug and drug-condition interactions against clinical databases. Not the theoretical pharmacokinetic interactions that clutter most reference tools, but clinically significant ones with management guidance. The kind of interaction that actually changes what you do.
Compare puts two or more agents side by side. Efficacy data, side effect profiles, dosing complexity, cost considerations. This is the tab for the moment you’re choosing between options and need to see them next to each other.
Switch references established medication-transition protocols. Cross-taper schedules, washout periods, equivalent dosing. This is one of the most common real-world prescribing tasks and one of the hardest to get right from a general AI query, because the answer depends on which direction you’re going and what the patient is currently stable on.
Indication works in the other direction. You start with a clinical problem and get back guideline-informed treatment options drawn from Canadian practice guidelines. Not a list of everything that’s ever been studied, but a guideline-informed view of what Canadian practice supports.
Deprescribe is built around the growing evidence on safely reducing medication burden, drawing on resources like deprescribing.org. For a family physician managing a patient on a dozen medications, the question of what to take away is clinically distinct from the question of what to add. It deserves its own tool.
Built on trusted Canadian references
We’ve partnered with the Canadian Pharmacists Association to bring CPS monograph data directly into Asklepius through a live API. This isn’t a model that read some drug information during training and now paraphrases it from memory. Drug Mode draws on CPS data at the point of query, and every response attributes the source it drew on.
Alongside the CPS, Drug Mode draws on Health Canada’s Drug Product Database — over 1,500 drugs — so the coverage extends across the products you actually prescribe.
This matters because these references are maintained and updated. Drug information changes. New safety warnings emerge, indications expand, dosing recommendations evolve. A model trained on last year’s data doesn’t know about this year’s changes. A system grounded in maintained reference data does.
When you ask Drug Mode about a medication, the information traces back to the same authoritative references you’d consult yourself. The difference is that Drug Mode structures that information around your specific clinical question, surfaces the parts that are relevant to the workflow you’re in, and makes it possible to do in thirty seconds what used to take five minutes of flipping pages.
Knowing what’s covered
Drug Mode now shows formulary coverage where available — whether a drug is covered, whether it’s restricted or requires special authorization, and the date the formulary data was last updated. Coverage varies by province and by plan, and it changes over time, which is exactly why the date stamp is there: it tells you how current the information is before you rely on it. The point isn’t to make the coverage call for you. It’s to put the relevant information in front of you at the moment you’re choosing what to prescribe.
How to use Drug Mode well
The biggest shift from a general AI tool is choosing the right tab before you type your question. The tab is the most important input you give the system, because it determines which data sources are queried, how the response is structured, and what clinical workflow the answer is optimized for.
A few patterns that get the most out of it:
Match the tab to your workflow, not the drug. If you’re choosing between two medications, go to Compare even if you know a lot about both drugs. If you need to transition a patient, go to Switch even if the drugs are in the same class. The tab tells Drug Mode what kind of answer to give you.
Give it the clinical context that changes the answer. Age, renal function, concurrent medications, indication. “Metformin dosing” gives you a generic answer. “Metformin dosing in a 72-year-old with eGFR 35 on lisinopril and empagliflozin” gives you a useful one. Keep it de-identified — clinical parameters, not patient identifiers.
Use Deprescribe proactively, not just reactively. Most physicians think about deprescribing when a patient has an adverse event. Drug Mode’s Deprescribe tab is designed to support the upstream question: which of these medications should I be reconsidering? It’s a tool for medication reviews, not just damage control.
Start with Info when you’re uncertain, then move to the specific tab. If you’re not sure which workflow you’re in, Info gives you the broad picture. From there you’ll usually know whether you need Interactions, Compare, or Switch.
What Drug Mode doesn’t do
Drug Mode is not a prescribing system. It gives you sourced clinical information to support your decision. It does not write prescriptions, it does not account for information it doesn’t have about your patient, and it does not replace the clinical judgment that comes from knowing your patient’s full history.
It also won’t always have a confident answer. Some drug questions land in territory where the evidence is thin, the guidelines are silent, or the clinical situation is complex enough that an AI tool shouldn’t be offering a definitive response. In those cases, Drug Mode will tell you what it can and flag what it can’t. And because Asklepius is built on Virtual Hallway, it can connect you to a specialist consultation when the question warrants it.
The CPS has always been a tool that supports physician decision-making. Drug Mode is built on that same principle. Better information, faster access, same expectation: you’re the one making the decision.