The science is already done
Trials, prices and patient counts sit in spreadsheets and source files.
Pramana by VedAGI, for Canadian market access teams
Drug companies already have the model and already use AI to write the submission. Pramana makes sure every number that leaves was computed from source, labeled honestly, signed by a person, and can be checked by the reviewer on the other side.
Pramana is Sanskrit for proof: a means of valid knowledge.
Example numbers only. Pramana blocks the pack and people fix it. It never quietly edits a figure.
The idea
Teams already build the model and already use AI to write the pack. The number then hides inside the writing, a person signs it anyway, and the other side cannot check it.
Pramana recalculates each number from source, marks guesses as guesses, refuses to release until a person signs, and gives the other side a receipt they can open in a browser. Same pack. Same model. The leak is closed.
The path
A company has a drug approved by Health Canada and wants public drug plans to pay for it. The numbers travel a long way before anyone is covered.
Trials, prices and patient counts sit in spreadsheets and source files.
A cost-effectiveness model and a budget impact model. Canada's Drug Agency asks for models it can trace, with no hidden hard-coding 3.
A reimbursement review application to Canada's Drug Agency, a French version for INESSS in Quebec, and later pCPA and provincial materials.
Often by an AI assistant now. The agency says any AI use must be declared, and the company stays accountable for every word 4.
Medical, regulatory and market access leads approve it and a named person signs. They trust the numbers match the model.
Reviewers at Canada's Drug Agency or INESSS often rerun the model and change the company's numbers 2.
pCPA negotiates a price and signs a letter of intent, which is only an option to fund. Each province then signs its own listing agreement 6.
On average it takes 906 days from Health Canada approval to a first public drug plan listing 1. Any number that has to be explained or fixed adds months to that path.
The gap
The weak joint is not whether AI can write. It is that whoever writes each sentence can copy, round, translate, or quietly change the number inside it. In Canada the same number is reused in several documents and two languages.
One figure lives in several documents and two languages. A hand edit, a translation or an AI rewrite changes one copy and not the others.
An uptake assumption reads like hard data because nothing marked it as an estimate.
The person signing cannot recompute every figure from source. The agency then reruns it and finds the gaps.
When the agency or pCPA asks where a number came from, teams dig through emails and old spreadsheets.
Canada's Drug Agency asks companies to declare how AI was used, keep a capable human in the loop, and stay accountable for everything in the submission 4.
The agency requires economic and budget impact models with a transparent trace, not hard-coded by hidden macros, and its reviewers rerun them 3.
When coverage claims are later reused in promotional pieces for health professionals, adding or changing them needs PAAB pre-clearance 5.
The industry already has models, writers and signers. It does not have a gate that says: this number was computed, labeled, signed, and can be checked by a stranger. That gate is the gap. Pramana is only that gate.
The gate
Keep the model. Keep the writer. Keep the reviewers. Pramana sits between “the words are drafted” and “the pack is signed and sent”, and it watches every copy in both languages.
Each figure is computed again, in code, from the company's own sources. The AI may suggest a formula. It never writes the value.
If a number is an assumption, it is marked as one in every document it appears in. It cannot pass as a measured fact.
The application, the French INESSS version, pCPA and provincial materials must all match the locked figure, or nothing is released.
What was computed, from which sources, with which formula, signed by whom, and where AI was used. Reviewers open it in a browser, with no VedAGI account.
One number
Follow a single figure, the year-one budget impact for public drug plans, from the company's files to a reviewer's screen at Canada's Drug Agency.
Pramana reads the company's own files: trial data, prices, patient counts and the model. Nothing new is invented.
The AI can propose how the number should be worked out. Pramana runs that formula itself, in code, on the real inputs.
The computed value is checked everywhere it appears, in English and French. Any mismatch is flagged, not quietly accepted.
Computed from measured sources, it becomes a figure. Built on an assumption, it stays a labeled estimate that everyone can see.
The signer approves computed figures, not paragraphs. Pramana records who signed, when, and what role AI played.
Value, formula, sources, signer and AI use, in one page any reviewer can open. It matches what the agency's AI statement asks companies to declare.
Locked. Anyone can recheck this number.
Example receipt. Values are illustrative.
Value
Pramana does not make a product look better. It saves the hours spent chasing numbers and protects the launch from avoidable slips. Anything we estimated is marked as an estimate, the same rule Pramana applies to its own numbers.
How we got this: About 40 key numbers, each repeated in about 4 places (agency application, French INESSS version, pCPA materials, summary), makes 160 spots. At about 10 minutes each by hand, that is about 27 hours per review round, times 3 rounds. Team time valued at C$150 to C$250 an hour.
906 days
average from Health Canada approval to a first public listing 1
Most detours come from real disagreements about value, which Pramana does not fix. It removes the avoidable ones: numbers nobody can trace, copies that disagree, guesses read as facts. For scale, average global drug sales run about US$500K a day 7, so one avoidable month is roughly US$15M pushed back (estimate).
| Question | Locked before it leaves | Corrected after someone else finds it |
|---|---|---|
| When the error is found | Inside the company, before sign-off | After the agency has already rerun your math |
| Who finds it | Pramana flags it automatically | Agency reviewers, INESSS, or pCPA negotiators |
| What it costs | A few hours to fix a flagged number | Clarification rounds, a weaker negotiating position, or a resubmission |
| What you can say | “Here is the receipt for every number.” | “We will look into it and get back to you.” |
We will measure, not guess. The first engagement locks one live results table and every copy of it, and records the real hours before and after. Those measured numbers replace every estimate here.
Who it helps
Each person who touches the number gets something different.
They approve computed figures, not paragraphs. If a number cannot be locked, it does not go out.
They keep the model they trust and stop chasing numbers across documents and languages.
On a 906-day path, a number that has to be explained or resubmitted costs months and weakens its position at the pCPA table.
They can check the receipt without buying anything. Trust moves from “we like this company” to “we can recompute this.”
A ready record of sources, formula, value, signer and AI use, in line with what the agency's AI statement asks for.
Limits
Pramana does not:
If the model is wrong, the lock will faithfully lock the wrong math. The benefit is integrity of release, not a better economic answer.
Start
Lock the budget impact results in one live Canadian submission, and every copy of it, including the French version. You keep your model and your review process. We measure the hours saved, and those measured numbers replace every estimate on this page.
Time and money figures marked as estimates are illustrative and will be replaced by measured results from the first engagements. Nothing on this page is legal or regulatory advice. Looking for the other market? Read the United States edition.
If you are about to send a table you cannot recompute, talk to us.
VedAGI. Intelligence for Humanity.