OPFOR Supply
OPFOR Supply plans investigational product for clinical trials: how many kits to make, when to make them and which depot should hold them. OnPoint Supply Solutions builds it for trial sponsors, contract research organizations and clinical supply teams running adaptive designs across global depot networks.
A planner enters the protocol, including arms, visit schedules, titration logic, kit definitions and depot lanes, and the workspace turns enrollment and site activation assumptions into kit demand by protocol, arm and visit rather than by quarter. Monte Carlo runs, at a stated 10,000 a minute, vary enrollment, titration and dropout and return P10, P50 and P90 bands around the production and distribution plan. Depot planning covers lane lead times, customs buffers, cold chain limits, label groups and lot expiry across 38 countries, and the forecast flags projected shortfalls by depot and month and recommends transfers. Reports from the interactive response system feed a reforecast every 24 hours.
The company states demand accuracy of ±0.4% per cohort, a 98.2% average service level and 25% less expired drug than a sponsor's plan of record, and one unnamed Phase 3 program saw $745,000 come out of its projected supply plan. OnPoint's team builds the first model and says it goes live in days. The license is free for internal planning, with unlimited seats, protocols and scenarios. The service takes aggregate planning data only and is not a validated GxP system, so randomization, dispensing and drug accountability stay in the trial's own systems.
Capability Axes
An AI Health Index grade measures what a buyer can verify from public sources on the date shown. It is not a rating of how good the product is. A vendor can build an excellent system and grade low on an axis because it publishes nothing an outsider can check. How grades read
Simulation drives the product. Planner supplied assumptions about enrollment, site activation, dropout and titration become kit demand by visit, Monte Carlo runs put a range around that demand, and depot allocation follows rules for label groups, release dates, expiry and buffer stock. The site sells this as AI driven supply planning.
The method itself is scenario modeling over inputs the study team sets, with no trained model or parameters fitted to past trials. Daily reforecasts change those inputs as IRT actuals arrive, and nothing indicates that any part of the engine learns from the reports. A sponsor gets most of the value, the visit level demand model and the depot rules, whether or not a learned component sits underneath.
Recommendations stay recommendations inside the tool. The forecast flags a projected shortage by depot and month and proposes transfers and production changes, while randomization and dispensing remain in the study's interactive response system and the team decides what actually ships. The license bars use for patient safety or regulated drug accountability decisions without independent human review and validation.
That review is a term the customer accepts, not a step the workflow enforces: nothing requires a planner to approve a recommendation before orders, transfers or lot allocations change. The site's "every visit covered" promise sits beside a method in which daily reforecasts adjust orders and transfers before deviations cause disruption, with no checkpoint described in between.
The forecast has several named moving parts. Monte Carlo simulation places P10, P50 and P90 bands around the plan, and a change to an enrollment assumption or site activation ramp flows through visit driven kit demand into regional supply and production. Buffer thresholds by kit and depot, label group matching and the handling of each lot's release and expiry month are all part of the depot model. Lane lead times come from the company's own lane survey covering 2020 to 2026, measured door to door with a stated deviation of about 2.1 hours.
The stated speed is 10,000 runs a minute. Which probability distributions drive the runs, how forecast parameters are set and how engine changes are released are not stated.
The forecasting engine is OnPoint Supply Solutions' own, and every calculation it runs is described as in house, with no outside model provider in the loop. Study teams supply the aggregate enrollment targets and kit schedules, and IRT reports supply the actuals. Microsoft Entra handles organizational sign in and asks only for a user's name and email under administrator consent, a role separate from the engine that projects inventory and stockouts.
The terms let OnPoint host and process planner inputs to run and improve the service but name no hosting provider or subprocessor, so the parties that store forecast inputs and simulation results are not identified.
Four figures headline the homepage: a 98.2% average service level, 25% less expired drug than the plan of record, 165 scenarios modeled and a longest study window of 6.5 years. None names the studies, sponsors or period behind it, or how the comparison with a plan of record was run.
The one worked example is a multiyear Phase 3 program in which a change the forecast caught took $745,000 out of the projected supply plan. That is a revision to planned spend, given without the sponsor, actual spend or costing baseline. The demand accuracy claim of ±0.4% per cohort has no method behind it either, and no independent study of shortages or wastage under the product has been reported.
Inputs are meant to be aggregate, such as enrollment targets, site counts and kit schedules. The terms of service, dated November 7, 2026, prohibit protected health information and any individually identifiable subject data, and users keep ownership of what they enter. OnPoint may host, store, process and display those inputs to run, maintain and improve the service, and commits not to sell them. The privacy policy, effective July 15, 2026, covers demo and account contacts and keeps personal information only as long as its stated purposes need.
How long trial inputs are kept, how the service would catch an identifiable record entered by mistake and whether account data can train anything are left open.
Protected health information is kept out by contract. The service is scoped to aggregate and deidentified planning data such as enrollment targets, site counts and kit schedules, and the terms expressly prohibit protected health information and identifiable patient or research subject data. Customers keep ownership of their inputs, authorize processing only to run and improve the service, and OnPoint commits not to sell them.
With no patient data in scope there is no business associate agreement tier, and the free license offers none. The website privacy policy covers people who request a demo or open an account, including SMS consent that is not shared with third parties. Compliance in practice rests on what the study team enters.
Access control is the visible security layer. Organizations on Microsoft Entra grant administrator consent for sign in, restrict access to assigned users or groups and can revoke the application for the whole organization, and sign in requests only name and email claims. The privacy policy refers to appropriate technical and organizational measures against unauthorized access, alteration, disclosure or destruction without naming any, such as encryption at rest or in transit.
OnPoint names no SOC 2 report, ISO 27001 certificate, penetration test or trust center for its own operations. The free license also disclaims uptime and continuity, so no service level stands behind the platform that holds a sponsor's supply plans.
OPFOR Supply sits outside the regulated record. The terms describe a planning aid for sponsors, contract research organizations and supply professionals that is not a validated computerized system under GxP or 21 CFR Part 11, and not a system of record for supply, drug accountability or regulatory submissions. The license bars use for direct patient care, patient safety or regulated accountability decisions without independent human review and validation.
The product models investigational product demand, allocation and potential shortfalls from assumptions the study team enters, while live randomization and dispensing stay in validated trial systems. A sponsor that uses a projection operationally takes on its validation, as the terms require.
Uncertainty is handled at the input level. Planners vary enrollment, dropout and titration, and the percentile bands show how far the supply plan moves as those assumptions move, which makes visible how risk compounds across country ramps and depot allocations. The homepage's accuracy figure of ±0.4% per cohort comes with no account of how it was measured or across which studies.
There is no stated process for testing whether forecasts hold up across protocol designs, regions or enrollment patterns, and no release review for changes to the simulation engine. The one worked example, a budget revision in an unnamed Phase 3 program, carries no breakdown by study type or geography.
A projection can be checked against its inputs. Planners rerun the forecast after changing a protocol assumption to see what drives a shortfall, and lot release and expiry dates, label groups and buffer policies give concrete values to inspect when a result looks wrong.
The terms put the burden on the sponsor. The user alone validates any output before relying on it, the software comes as is with no uptime, support or continuity commitment, and losses such as drug shortages, shipment delays and study delays are excluded. Aggregate liability is capped at the greater of $100 or the fees paid in the prior 12 months, which on a free license means $100. No remedy is offered for a supply decision made on an erroneous projection.
IRT is the one named connection. Reports from the trial's interactive response system feed Actuals Reforecast on a 24 hour cycle, pushing current enrollment into kit projections and a daily view of regional depot coverage, so the plan of record can be set against actuals as the study runs.
The boundary is deliberate: the tool plans supply and the operational systems randomize and dispense. Which IRT vendors it reads, whether reports arrive by file, API or manual upload, and whether anything writes back to the IRT are not stated. Hospital records fall outside a planning tool that works on aggregate data, and no other trial system, such as a CTMS or ERP, is named.
Most customers use the hosted web workspace. Organizations on Microsoft Entra can sign in through it, limit access to assigned users or groups and revoke the application centrally. A dedicated tenant, private deployment or self hosted instance is possible under a separate agreement at OnPoint's discretion.
For the standard service, the terms name no facility, cloud provider or region for storage and processing, and say nothing about where backups sit or how one customer's data is kept apart from another's. Maine law governs the agreement, with disputes heard in Oxford County, which places the company but not the data. Any location commitment would come in the separate agreement, which can carry its own data handling terms.
The standard license is free for internal trial planning, with unlimited seats, protocols and scenario runs and no fee per study or per user. Prospects can sign in and build a forecast on their own study or book a demo, and the team offers to build a sponsor's first portfolio model.
Paid tiers or usage limits may arrive later, applied prospectively, and the free license carries no uptime or support commitment. Dedicated tenancy, private deployment and self hosted instances need a separate agreement with its own fees and service terms, and none of those prices are listed. Nor is any charge for the assisted model build, which the site presents as part of a turnkey setup.
Clinical supply teams at sponsors and contract research organizations are the buyers, planning production and distribution for adaptive studies with multiple arms, titration branches and regional depots. The depot model handles label groups, lot expiry, lane lead times, customs buffers and cold chain limits across 38 countries, cross study aggregation supports long range portfolio plans, and the longest study window modeled so far runs 6.5 years.
No sponsor, therapeutic area or mix of study phases is named beyond the unnamed Phase 3 example, and demand accuracy is not broken out by design or region. Device trials, and studies that ship directly to patients' homes, are not mentioned either way.
Alternatives to OPFOR Supply
Other products that do the same kind of work as OPFOR Supply, and what each one does.
PhaseV provides clinical development software for pharmaceutical sponsors, biotechnology companies and contract research organizations, from offices in the Boston area and Tel Aviv, and says more than 40 biopharma companies and CROs use it.
Clinical trial platform that embeds patient identification, eSource and data capture inside provider organizations so trials can run in routine care, paired with its own recruitment coordinators and eligibility specialists.
QuantHealth is a clinical trial simulation company with offices in Tel Aviv and Worcester, Massachusetts.
ConcertAI's Accelerated Clinical Trials platform helps pharmaceutical sponsors and contract research organizations design, plan and run oncology studies.
Unlearn, based in San Francisco, builds Digital Twin Generators: disease specific machine learning models trained on historical patient level data that forecast how each clinical trial participant would progress under control or standard of care, from their baseline data.
Medidata, a Dassault Systemes brand based in New York, sells clinical trial software to drug and device sponsors and contract research organizations.
Pricing
Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.
| Entry Price | Pricing Basis | BAA Tier | Implementation | Source |
|---|---|---|---|---|
|
$0 for internal trial planning
$0 baseline
|
The standard license is free for internal trial planning. The hosted workspace includes unlimited seats, study protocols and scenario runs with no per study fee. Dedicated tenancy, private deployment and self hosted arrangements require separate terms. | The service accepts aggregate and deidentified planning data only. Its terms prohibit identifiable study participant information and protected health information. The standard license has no business associate agreement tier. | OnPoint's team builds the first portfolio model as part of a turnkey setup, with no fee stated for that work. The free license makes no service level commitment, and fees for dedicated or private deployment are negotiated. | Vendor Published |
The terms of service and license, both dated November 7, 2026, allow paid tiers, usage limits or new features to be introduced prospectively. A separate written agreement for a dedicated tenant or private deployment can carry different fees, support commitments, data handling provisions and liability limits.