PRISM turns 35 years of pricing strategy into one auditable answer: your probability of winning at every price, calibrated on your history, stress-tested against your competitors' behavior, and traceable down to every source.
In hospital and vaccine markets, one tender can carry a full season of volume. The price you write on that page is the highest-stakes decision of the year, and it is usually taken with a spreadsheet, an anecdote, and a deadline.
Competitors appear and vanish based on capacity, financial pressure and strategic intent. A rival with an empty plant does not behave like a rival with a full one.
A competitor under financial strain, or with a vital need to win a reference market, will bid below any historical pattern. Reading that intent is the game.
Lowest price, weighted quality scoring, split awards: each mechanism rewards a different strategy, and each country plays it differently.
For every price you could bid, the engine simulates thousands of editions of your tender: who shows up, how they price, who wins under the actual award mechanism. The result is not an opinion. It is a counted frequency, with an explicit margin of error and a recommended price corridor where expected value peaks.
No AI ever computes a probability in PRISM. Language models prepare and justify assumptions, your experts validate them, and a deterministic engine does the mathematics. Every figure can be reproduced bit for bit, months later, in front of any committee.
Three stages, with a hard gate in the middle. That gate is not a safety disclaimer, it is the product's core mechanism: your organisation's judgment enters the model in a form that is explicit, structured and journalled.
AI agents read your tender history, award notices and competitor material. They propose every assumption with a written justification and the exact source excerpt behind it. They never guess: gaps are flagged, not filled.
AI preparesIn the parameter room, each assumption is read, then validated or overridden with a motivated reason. Nothing runs until 100% is signed off. Your team's tacit knowledge enters the model here, traced and timestamped.
You decideTen thousand simulated editions per run. The engine applies the tender's real award mechanism, prices in international reference price contagion, and produces the curve, sensitivities and war-game scenarios.
Mathematics computesMost AI pricing tools fail at the boardroom door because nobody can explain their numbers. PRISM was engineered in the opposite direction: the explanation is the product. Rationale first, percentage last.
This is not a video. The panel below runs the same simulation logic as PRISM's engine, in your browser, on a fictional Nordic vaccine tender of 1.2 million doses. Move the sliders and watch the curve answer.
Live Monte Carlo, 3,000 editions per slider move, seeded and reproducible: the same inputs always return the same curve. Fictional franchise, fictional competitors, illustrative figures. In the platform, every assumption behind these curves is human-validated and evidence-linked before anything runs.
Notice how the value-optimal corridor moves when competitor pressure rises: the right answer is rarely "bid lower everywhere", it is a precise repositioning, priced against the risk of international reference price contagion. That reading, not the software, is what CVA brings to your table.
Ask any black-box tool what it actually takes into account and the conversation ends. In PRISM it is a screen. Seven families of drivers feed every probability, each one stored, sourced, validated by your team and cited by name in the audit trail.
Plant-level capacity per competitor, evidence-linked: the denominator of every feasibility judgment the model makes
Volumes already committed from previous tender awards, tracked season by season: a full plant does not bid like an empty one
Every bid observed elsewhere anchors the likely-bid ranges, weighted by recency and similarity, with the source excerpt attached
A rival's behavior on other vaccines reads through to yours: typed, dated signals captured across the portfolio, not just the tender at hand
Sales-force chit-chat and market whispers enter as named, human-attributed signals, graded for reliability, and never sufficient on their own
A challenger who needs the cash does not bid like an incumbent defending margin on a non-core line: archetypes and need-to-win scores, argued and sourced
Established player, challenger or new entrant, core or secondary market for them: position shifts participation and aggressiveness through published rules
If it is not in the registry, it is not in the number. And the registry shows its gaps: every driver carries a coverage status per competitor, so you always know what the model knows, and what it honestly does not.
The sharpest tender play is sometimes a deliberate loss. Concede a low-margin reference tender, let a rival fill their plant, keep the price war away from your reference price, and watch your odds rise on the bigger tenders downstream. PRISM simulates the season sequentially: every award consumes competitor capacity and reshapes the odds of everything that follows.
Toggle the strategy above: conceding the reference tender fills the rival's plant, protects your reference price from IRP contagion, and lifts your odds on the tender that actually matters. Sometimes the numbers say the opposite. PRISM's job is to price the trade-off, not to romanticize it.
Season play is where tender pricing stops being a procurement exercise and becomes portfolio strategy: which markets to defend, which battles to concede, where a rival's full plant is worth more to you than a marginal win. That judgment is CVA's terrain, and PRISM makes it quantifiable.
Sophistication is not reliability. PRISM is built on a harder discipline: calibration. After every awarded tender, the real outcome is recorded and confronted with what the model predicted. Over time, the platform publishes its own track record: when it says 70%, does it win about 70% of the time?
Assumptions start from measured frequencies in your own tender record, never from generic industry averages, then adjust for documented competitor behavior through transparent, published modulation rules
Wide evidence, narrow ranges. Thin evidence, wide ranges and a visible low-confidence flag. The platform never manufactures false precision, and its confidence bands say so on screen
Every debrief sharpens the behavioral rules. The system becomes more accurate with every tender cycle, turning your tender history into a proprietary strategic asset competitors cannot copy
Pricing intelligence is among the most sensitive data a pharmaceutical company holds. The platform's architecture starts from that fact.
Each client runs a single-tenant instance on European infrastructure, isolated by design. Deployment inside your own cloud environment is available on request
Client data is used to calibrate your instance and nothing else. It is never used to train AI models, never pooled across clients, never leaves your perimeter
Sources, assumptions and overrides are journalled in a ledger that can be extended but never silently edited. Who changed what, when and why is always answerable
Every simulation is seeded and archived with its full manifest. Any figure shown to a committee can be independently recomputed and verified, months later, to the last decimal
Every session is tied to a named user and a role. Validations and overrides carry their author. Access is granted per franchise and per market, and revoked centrally
Franchise and country workspaces are compartmented, supporting internal confidentiality walls and, where relevant, clean-team setups during competitive processes
PRISM is deployed the way any decisive strategy engagement is run: senior-led, evidence-based, and built around your teams. The deliverable is not a licence, it is a decision capability your organisation owns, with CVA at your side when the stakes peak.
Anyone can sell you a dashboard. The value is in the assumptions, and assumptions are strategy work: competitor intent, buyer behavior, reference-price contagion, portfolio trade-offs. That is what our consultants have done for 35 years, and the platform is how that work now compounds instead of evaporating after each engagement.
"A probability is worth exactly the rigor of the assumptions behind it. For 35 years, our work has been the assumptions. Now they compound."CVA New Healthcare Systems, Corporate Value Associates