Scenarios
Test the decision before you make it
The simulation runs on your own milestones, sites, batches, budgets and dossier, not on an industry average. Results are ranges, and every assumption is recorded.
Engines
Seven simulation engines
Each engine uses only your records and assumptions you can see. Run it again with the same inputs and you get the same result. Every forecast is later checked against what actually happened.
| Engine | Model | Headline outputs |
|---|---|---|
| Timeline | Dependency graph over playbook steps, tasks, milestones and submissions; slip distributions propagated in order | Completion date per milestone, on-time probability, critical path frequency |
| Runway | Monthly cash with burn drift, milestone-driven costs, grants with award probability and a planned raise | Months of runway, cash-out date, cash at 12 and 24 months, funded probability per milestone |
| Enrollment | Gamma-Poisson site model with startup times, added sites, competing trials, dropout and screen failure | Last patient in, readout, months to target, evaluable N |
| Statistical design | Closed-form sample size and power for continuous, binary and time-to-event endpoints; placebo erosion; assurance | Required N, power at planned N, probability of success, minimal detectable effect |
| Supply | Monthly stock by batch with expiry, demand from enrollment, reorder policy, yield, failure and lead times | Stockout probability, first stockout month, expired units, batches ordered |
| Regulatory pathway | 505(b)(1), 505(b)(2), IDE, EU CTA and 510(k) templates with meeting lead times, dossier completion, statutory clocks and hold branches | Submitted date, study-may-proceed date, hold probability, on-time probability |
| Outcome priors | Elicited effect, adverse event and adherence priors propagated through the design to a readout | Positive readout probability, safety-stop probability, observed effect |
Workflow
How an experiment runs
Each experiment runs as a sequence of steps. If a step fails, it resumes where it stopped, and no step runs twice. Without an AI model key, the experiment still completes using default and record-derived assumptions.
- 01
Snapshot
A copy of your records is saved and identified by its content. Runs on the same copy give identical results.
- 02
Evidence and assumptions
Comparable trials, papers and adverse event data are pulled from ClinicalTrials.gov, PubMed and openFDA. Assumptions are set from that evidence, each with a confidence level and citations.
- 03
Scenarios
The options you want to test become scenarios: more sites, a later raise, a different pathway or endpoint, a rescheduled batch.
- 04
Engines
Each engine runs for each scenario with the same seed and time horizon. Every output is a range or a probability.
- 05
Ranking and panel
Scenarios are ranked against your constraints. Expert reviewer personas, including an agency reviewer in the war game, return concerns and questions.
- 06
Report
A credibility record and a report are written into the Data Room as a versioned, signable document.
Credibility
Each result records its assumptions and limits
Every experiment includes a credibility record that follows ICH M15 and the FDA AI credibility framework. It states the question, the intended use, the consequence of the decision, how much depends on the model, the model risk, the engines and versions used, where each assumption came from, how the results were verified and validated, and the limitations.
The platform does not predict clinical efficacy or regulatory approval, and the interface never shows a single-number estimate of either.
Learning loop
Forecasts are checked against what happens
Each run predicts milestone dates, site activations, enrollment counts, batch depletion and the cash balance at 12 months. Every night the platform compares open forecasts with the actual records, scores them (CRPS, Brier score, 80% interval coverage and bias in days) and adjusts the engines for any measured bias. When records change materially, it re-runs the stress test and files a risk for each critical finding.
In the product
What an experiment looks like
Start from a template
Timeline shock, runway stress, enrollment and design, endpoint selection, supply check, regulatory pathway, war game or a company stress test.
Ranges over time
Every series shows the median per scenario with the P10 to P90 band of the baseline.
Findings with severity
What each engine flagged, per scenario, with the run that produced it.
Priors with their evidence
Each assumption shows how it was set, its confidence level and the trials or papers it cites.
Every run can be inspected
Metrics with P10, P50, P90 and mean, per-site tables, and the exact assumptions the engine saw.
Forecasts with resolution dates
Predictions about your records, scored automatically when the actual value is known.
Regulatory war game
An agency reviewer persona reads the dossier documents and returns the questions you should prepare for.
Reports in the Data Room
Versioned, reviewable and signable like every other document.
Agent environment
A sequence of decisions on a copy of your records, scored by the engines. Make the decisions yourself or compare Claude against simple baseline strategies.
See it with your own program
We start the demo on a sample company, then use your own records. Bring a decision you are currently considering and we will simulate it during the call.