CLINICAL FORECASTING
Clinical Forecasting, Operated by Agents
Forecast trial spend, reconcile CRO invoices, and produce audit-ready clinical accruals — all driven by AI agents trained on biotech finance. The way it should have worked in spreadsheets, but never did.
THE STAKES
Clinical budgeting and forecasting drive 70% of biotech R&D spend — and most of it still lives in spreadsheets.
For a clinical-stage biotech, the clinical budget is the budget. Phase II trials average $60M. Phase III trials average $300M and routinely exceed $1B. Across the industry, clinical development accounts for 60–70% of total R&D spend. Get a clinical trial forecast wrong by 20% — which is the average overrun — and you've blown a $60M hole in your runway, missed a board commitment, and put your next financing round at risk.
The numbers behind the discipline are unforgiving. Roughly 80% of clinical trials run late. CRO invoices lag actual trial activity by 60–180 days, which is the structural reason clinical accruals are so hard to estimate. And in 2025, PwC reported that 67% of pre-commercial biotechs still manage the CTMS-to-accounting integration manually in Excel. So the highest-stakes workflow in biotech finance — the one that determines runway, audit risk, and board credibility — is also the one most likely to be running on copy-paste and named ranges.
Existing options have not closed the gap. Spreadsheets break the moment a protocol amends or a CRO changes invoice format. Legacy clinical FP&A platforms force you to model your trial the way their rule engine expects, behind logic you cannot audit or extend. And generic FP&A tools — Adaptive, Anaplan, Pigment — have no clinical primitives at all. They cannot model patient-level activity, site activation curves, or hybrid CRO contracts. So biotech finance teams end up bridging the gap with their own discipline: a heroic Excel workbook, weekly CRO calls, and a quarterly accrual review that is more art than audit.
Exigyn's clinical module solves this by inverting the architecture. Instead of a closed platform you configure, it is an open agent framework you extend — built on two foundations that did not exist when the legacy platforms were designed.
Built on the Model Context Protocol.
The Exigyn clinical module is built on the Model Context Protocol (MCP) — the open standard Anthropic introduced in late 2024 and donated to the Linux Foundation's Agentic AI Foundation in December 2025. MCP is now the de facto way AI agents connect to data sources in production, with thousands of active server implementations across enterprise, finance, and life sciences.
For a biotech finance team, that translates into three concrete properties spreadsheets and closed SaaS cannot match. First, your data stays in your stack: MCP servers run inside your environment, so CTMS, EDC, ERP, and CRO portal data flow into agents securely and ephemerally — no SaaS vendor moonlighting as a data lake. Second, every agent action is logged at the protocol layer, which means when your auditor asks "how did the agent compute this $2.4M clinical accrual?", the answer is reproducible from first principles instead of a black-box model. Third, MCP is an open spec, so the data and agents you stand up today are not trapped in a proprietary format tomorrow.
Customizable through Skills.
On top of MCP, every Exigyn customer authors Skills — reusable, version-controlled bundles of methodology, prompts, and reference data that the clinical agents load on demand. A Skill might be your firm's specific clinical accrual methodology, your auditor's preferred backup format, your CRO's contract template, or your board's EAC reporting standard. The Trial Forecast Agent and Accrual Agent read Skills the same way a senior FP&A analyst reads a methodology memo on day one.
This matters because clinical finance is bespoke. A Series A oncology biotech and a public Phase III cardiovascular biotech run entirely different trial economics — different unit costs, different milestone structures, different passthrough markups, different audit standards. Closed platforms force you to flatten that complexity into their model. Exigyn's clinical module loads your Skills, and the agents conform to the way your finance team thinks. The result is a clinical forecasting and accrual system that is live instead of point-in-time, continuous instead of quarterly, auditable instead of opaque, and yours instead of the vendor's.
THE PROBLEM
Clinical forecasting is broken in spreadsheets — and rigid in legacy SaaS.
Clinical trial forecasting is the highest-leverage workflow in biotech finance. It drives cash runway, board confidence, and the accuracy of every clinical accrual you book. But the way most clinical-stage biotechs do it today is broken in three specific ways.
Spreadsheets break the moment a protocol changes. A typical trial budget links to vendor schedules, site activation timing, patient enrollment curves, and CRO unit-of-work pricing. Change one assumption — protocol amendment, enrollment delay, vendor change order — and you're rebuilding cells across twenty tabs. Most teams give up forecasting trials in detail and fall back to straight-line monthly burn assumptions that drift further from reality every quarter.
Legacy clinical FP&A platforms solve part of this but bolt their data model onto your trial. You configure the platform to match how it expects clinical operations to work, not the other way around. Onboarding takes 8–12 weeks. Modeling assumptions live behind opaque rule engines you can't audit.
ERP-attached planning tools handle the financial structure but lack any clinical primitives. They know how to model accounts and departments — they don't know what a per-patient-per-visit cost is. Your team ends up exporting CRO spend from one system, reconciling it in Excel, and pasting a one-line accrual entry back into the planning tool. The audit trail is broken before the accrual is even booked.
HOW IT WORKS
Agents handle the work spreadsheets can't and SaaS won't.
Exigyn's clinical forecasting workflow runs on a small set of agents you can audit, not a black box you can't.
The Trial Forecast Agent ingests your protocol, enrollment curve, vendor contracts, and CRO unit pricing. It builds a patient-level forecast: cost per patient per visit, multiplied by enrollment trajectory, plus passthroughs and milestone-driven payments. When you change an input — delayed FPI, lower screen-fail rate, new site activation — the agent recalculates the forecast and explains every line that moved.
The Accrual Agent reconciles vendor invoices to the forecast every month-end. It flags variance, surfaces unusual line items, and produces an audit-ready clinical accrual entry with full backup. ASC 730 compliant. No more straight-lining; no more month-13 catch-up adjustments.
The Variance Agent watches actuals vs. forecast continuously, not just at quarterly close. It posts to your finance Slack when a metric drifts beyond a threshold you set, and proposes a forecast revision you can accept, reject, or amend.
All three agents work against the same underlying clinical model — protocol, enrollment, vendors, milestones, unit pricing — so when reality diverges from the plan, every downstream artifact updates together. Your forecast, your accrual, your board pack, and your runway all stay in sync.
CAPABILITIES
Built for the way biotech clinical finance actually works.
Every primitive a clinical-stage biotech finance team needs — without the eight-week implementation.
— Patient-level forecasting with enrollment curve and dropout assumptions
— Vendor-level spend modeling: CRO, central lab, IRB, IxRS, imaging core lab
— Milestone-driven and unit-of-work pricing engines (per-patient, per-visit, per-site, per-monitoring-day)
— Site activation timing and ramp curves
— Country mix and passthrough modeling for global trials
— Protocol amendment scenario branches with side-by-side diffs
— Audit-ready clinical trial accruals (ASC 730 compliant)
— Live CRO invoice reconciliation
— ROFO, PASS, and EAC roll-forward for board reporting
— Variance attribution split by protocol, vendor, and operational drivers
— Cash conversion modeling: when forecast spend turns into invoiced cash out
— Live integration with CTMS, EDC, and ERP systems
METHODOLOGY
How Exigyn calculates clinical trial accruals.
Clinical trial accruals are where bad forecasting becomes a financial reporting problem. The clinical accrual you book each month should reflect economic reality — services delivered but not yet invoiced — not vendor billing cycles. Done badly, you get hockey-stick adjustments at audit. Done well, your reported R&D burn is a true and timely picture of trial economics.
Exigyn calculates clinical accruals using a unit-of-work methodology tied to actual trial activity, not vendor invoice timing.
For patient-level activity, the Accrual Agent uses your enrollment data — pulled from CTMS, EDC, or manual upload — and your vendor contract pricing to compute services delivered in the period. Each patient visit, screen, randomization, or treatment cycle that has occurred but not yet been invoiced is recognized in the right month. This is the foundation of patient-level accrual: tying the accrual to what actually happened in the trial, not to when the CRO chose to send a bill.
For site-level activity, the agent recognizes site-month fees, passthroughs, and activation milestones based on actual site status — not invoice timing. Site activation accruals follow the milestone schedule in the contract; site maintenance fees follow the activation status reported by the CRO each month.
For CRO services with hybrid pricing — which most modern CRO contracts use — the agent splits the contract into components and accrues each appropriately. Unit-of-work components (per-patient, per-visit) accrue based on activity. Milestone components accrue against the milestone schedule. Passthroughs flow through at the rate the CRO marks them up by, with full visibility into the markup percentage.
The output is an audit-ready clinical trial accrual entry with full lineage from each line item back to the underlying clinical activity, vendor contract, and unit price. When your auditor asks "what's behind this $2.4M accrual?" — every dollar is one click away from the patient visit, site activity, or milestone that justifies it.
This methodology aligns with ASC 730 (Research and Development Costs) and supports both 404a and 404b filers. It also gives your CFO and audit committee something they almost never get from spreadsheet-based clinical accruals: confidence that the number reflects what the trial is actually costing, not what your CRO has remembered to bill.
HOW WE COMPARE
Clinical forecasting head-to-head: Exigyn vs the legacy platforms.
We did the comparison work for you — Exigyn against Auxilius, Condor, Medidata CTFM, IQVIA CTFS, Adaptive, and Pigment, line by line, on the dimensions that actually matter for clinical-stage biotech finance teams. No feature theater — just what each vendor actually ships in the base plan.
TRY IT
Try clinical forecasting in the live studio.
Skip the demo call. The studio is preloaded with a fictional clinical-stage oncology biotech — cap table, trial budget, enrollment curve, and CRO contracts. Adjust an input. Watch the forecast and the clinical accrual recalculate together.
FAQ
Clinical forecasting questions, answered.
What's the difference between clinical trial forecasting and clinical trial budgeting?
A budget is a fixed, point-in-time agreement on what a trial should cost. A forecast is a continuously-updated estimate of what the trial will actually cost, given everything you've learned since the budget was struck. Most clinical-stage biotechs build a budget once and never re-forecast in detail. That's where the gap between "what we said it would cost" and "what we're actually spending" opens up. Exigyn's Trial Forecast Agent keeps the forecast live against the budget, with variance attribution by protocol, vendor, and operational driver.
How does Exigyn integrate with our CTMS and EDC?
Exigyn pulls patient-level activity (enrollments, screens, visits, dropouts) from common CTMS and EDC systems via API or scheduled export. The Trial Forecast Agent uses this data to drive both the forward forecast and the unit-of-work clinical accrual calculation. If you're on a system without a clean API, monthly file uploads work too.
Are Exigyn's clinical accruals ASC 730 compliant?
Yes. The Accrual Agent calculates clinical accruals using a unit-of-work methodology that ties accruals to actual trial activity (patient visits, site activations, milestones) rather than vendor invoice timing. Output includes full audit lineage from accrual line item back to the underlying clinical activity, vendor contract, and unit price. Supports both 404a and 404b filers.
How does Exigyn handle protocol amendments?
Protocol amendments create a scenario branch in the underlying clinical model. You can run the amended protocol alongside the original as scenarios, see the forecast and accrual impact side-by-side, and promote the amended version to baseline when the protocol amendment is finalized. Historical accruals on the prior protocol are preserved for audit; forward accruals run on the amended protocol from the effective date.
Can we use Exigyn alongside Adaptive, Anaplan, or NetSuite?
Yes. Exigyn handles the clinical-specific layer — trial forecasts, CRO spend, clinical accruals — and pushes summarized output to your existing FP&A and ERP systems. The clinical accrual entry posts to NetSuite (or your GL of choice) with the same lineage and backup that auditors require.
What's the difference between Exigyn and Auxilius or Condor?
Auxilius and Condor are excellent at the clinical accrual workflow specifically. Exigyn is broader: it covers the full forecast → accrual → variance loop, plus the rest of biotech FP&A (cash runway, headcount, OpEx) on the same data model. The architecture is also different — Exigyn is built on agents you can audit and customize, not a closed rule engine. Our /compare page walks through the differences in detail.
How long does it take to onboard a trial?
48 hours to a working environment with your data loaded. Two weeks to full agent calibration against your existing forecast and accrual methodology. No 8–12 week implementation. No per-trial onboarding fees.
Who owns the model — Exigyn or our finance team?
Your finance team. Every assumption, formula, and agent prompt lives in your workspace. You can read every line of every agent's logic, change it, and roll back versions. Exigyn provides the platform and the agent primitives; the model is yours.
GET STARTED
Your clinical forecast. Live in 48 hours.
Run a 30-day pilot for $500, credited to your first month. Preloaded oncology pipeline, import your own data whenever you're ready.