Schedule risk management for biotech programs

Drug development programs fail their schedules in characteristic ways, and none of them are solved by adding more columns to a spreadsheet. The estimates are uncertain (a tox study read-out, a process-development run, an analytical method qualification), the resources are constrained (a single CMC vendor, one regulatory lead, a shared GMP slot), and the gates are real (you either advance the candidate or you do not).

Against that, the cost of being wrong is brutal. In late-stage pharma the cost of delay runs on the order of $800K per day, so a single quarter of IND acceleration can be worth tens of millions. The schedule is not a planning artifact — it is the single largest financial lever a development organization controls.

CritPath runs Monte Carlo simulation over your actual schedule network — PERT-Beta task distributions, risk-event injection for known threats (a failed batch, a clinical-hold scenario), optional duration correlation, and a criticality index that tells you which activities most often land on the critical path across thousands of runs.

The output is the number your board deck has been missing: P50, P80, and P90 finish dates for the IND filing, the FIH-enabling tox package, or any milestone you choose. Instead of committing to a median everyone privately discounts, you can commit to a P80 you can defend — and show the tornado chart that explains which uncertainties drive the tail.

Stage gates are first-class objects in CritPath, not calendar reminders. A gate carries an explicit Go / No-Go / Pivot / Defer decision, and the decision is wired into the schedule. When you make the call at a candidate-selection or IND-readiness gate, the downstream plan responds to it rather than sitting frozen while the PMO rebuilds the Gantt by hand.

That turns a gate review from a slide into a model. You can see what advancing the program does to the resource-constrained critical chain, what deferring buys you in buffer, and what a pivot costs in elapsed time — with the probabilistic dates updating underneath the decision.

This is the day that used to cost a day. A contract manufacturer pushes a GMP campaign two weeks, or a CRO read-out moves, and historically that means an afternoon of dragging bars and a follow-up email to everyone whose dates just shifted.

In CritPath, you log the slip once and the engine re-cascades the whole plan: the Critical Path Method recomputes float and the near-critical set, the critical chain and feeding buffers re-absorb what they can, and the Monte Carlo distribution refreshes so you immediately see whether your P80 IND date actually moved or whether buffer absorbed it. One input, a fully consistent re-plan, and a fever chart that shows how much protective buffer you just spent.

CritPath ships AACE RP 132R-23 Level 4 risk-driven scheduling with an append-only audit log, so the provenance of every estimate, risk, gate decision, and schedule change is recorded. For a development organization that will eventually face diligence, an acquirer's data room, or a regulator, the schedule stops being a black box and becomes an evidenced record.

To be precise about what is and is not shipped: the AACE L4 methodology and the audit-trail foundation are live today. 21 CFR Part 11 electronic-signature attestation, SOC 2 Type II, HIPAA BAAs, SAML SSO, and on-prem deployment are on the Enterprise roadmap, not yet shipped — we would rather under-claim than have a pharma buyer discover a gap in qualification.

The differentiator legacy desktop tools cannot match is the copilot. Built on Claude and Gemini and grounded in your live dependency graph, it answers the question a program leader actually asks: "which task is driving my P80 slip, and what happens to the IND date if the tox study runs two weeks long?" It reasons over the real engine — CPM, Monte Carlo, buffers, gates — not a chatbot pinned on top of a generic tool.

Around it sits the full method stack a biotech program needs in one place: CPM with four dependency types, lag, float and near-critical analysis; CCPM with Theory of Constraints and Drum-Buffer-Rope (project and feeding buffers, fever chart); WSJF and Cost of Delay to prioritize across a portfolio; EVM (SPI/CPI/EAC) for funder and board reporting; configurable AI employees with custom skill sets; multi-tenant orgs, 2FA, an append-only audit log, and XLSX/CSV import with AI work-breakdown decompose to get an existing plan in fast. During the beta, autonomous AI-agent runs are off; AI assistance — the copilot, decompose, reporting, and skill wizards — is on.

Import an existing program from XLSX or CSV (or let the AI decompose a one-line scope into a work breakdown), set three-point estimates on the activities that matter, mark your gates, and run the simulation. Within an afternoon you have a P80 IND date with a tornado chart behind it instead of a deterministic line.

CritPath is $10 per user per month for every standard feature and unlimited projects, with AI copilot usage billed separately by metered token usage and Enterprise available as a custom plan. That is a fraction of a single legacy schedule-risk seat — and it runs in a browser the whole program team can open, not on one analyst's desktop.