Book a demo
Menu

How it works

How Provenesis models, plans, and explains.

This page is for the people who own the plan: supply chain, manufacturing operations, planning, and tech ops. It covers what Provenesis represents, the decisions it makes, and what it gives back.

Book a demo

01The network model

Your network as it actually runs.

Provenesis represents the whole network, not one site at a time: where patients are collected, how material moves, where it is manufactured and released, and where it is infused.

Each element carries its real limits: the slots in a suite, the operators who run them, QC capacity by test, vector lots and their release dates, and which sites can supply which markets.

Collection
Apheresis sites and treatment centers, with their collection days and capacity.
Logistics
Courier lanes and pickup windows, fresh and cryopreserved material, and dry shipper availability.
Manufacturing
Sites, suites, and slots, with operators, shifts, and the steps of each process.
Release and supply
QC capacity by test, vector and critical materials, and which markets each site can supply.
An advanced therapy network and its decision pointsFour treatment centers on the left collect apheresis material and later infuse patients. Courier lanes carry material from the centers to two manufacturing sites, including one alternate lane. Site 1 has four suites and site 2 has three; each suite is divided into slots, some filled. Each site has a QC release area. Vector supply on the right feeds both sites. Released, cryopreserved product returns along a loop at the bottom to the treatment centers. Six numbered markers show decisions: collection date and site, courier lane and window, site, suite, and slot, QC release, shipment and infusion date, and vector allocation.TREATMENT CENTERSCOURIER LANESMANUFACTURING SITESVECTOR SUPPLYRELEASED PRODUCT, CRYOPRESERVEDapheresis material,fresh or cryopreservedalternateVECTOR LOTSlentiviral or retroviralrelease dates, lot sizesCENTER Acollect, infuseCENTER Bcollect, infuseCENTER Ccollect, infuseCENTER Dcollect, infuseSITE 1suites and slotsS1S2S3S4QC RELEASEanalysts, testsSITE 2suites and slotsS1S2S3QC RELEASEanalysts, tests123456
Fig. 2The network as Provenesis sees it. Apheresis material flows from collection sites to manufacturing, through QC release, and back to treatment centers. Numbered markers show where Provenesis makes decisions.
  1. Collection. Date and apheresis site for each patient.
  2. Courier. Lane and pickup window for the apheresis material.
  3. Manufacturing. Site, suite, and slot for each batch.
  4. QC release. Testing workload, analyst capacity, and the release date.
  5. Infusion. Shipment and infusion date, timed with lymphodepletion.
  6. Vector and materials. Lot allocation across sites and products.

02The patient journey

One patient's path, from vein to vein.

For an autologous therapy, every batch belongs to one patient. Provenesis plans each patient's whole path: collection, transit, activation, transduction, expansion, cryopreservation, QC release, transit, and infusion.

Chain of identity and chain of custody stay with your orchestration and quality systems. Provenesis plans around them: each line in the plan is one patient's material, from collection to infusion.

Collection scheduling
Which apheresis date and site, given the capacity waiting downstream.
Slot assignment
Which manufacturing site, suite, and slot for each patient's batch.
Release planning
QC workload and analyst capacity, so release dates are realistic.
Infusion readiness
A release date the treatment center can plan lymphodepletion and infusion around.
An autologous CAR-T patient journey, from vein to veinTen stages from left to right: leukapheresis at the treatment center, transit of fresh or cryopreserved material, then at the manufacturing site activation, transduction with a lentiviral or retroviral vector, expansion, formulation, cryopreservation, and QC release testing, then transit in a cryogenic shipper, and infusion at the treatment center after lymphodepletion. A chain of identity line runs under every stage, and a bracket marks vein-to-vein time. Five numbered markers show decisions: collection date and site, courier window, manufacturing slot, release date, and infusion date.TREATMENT CENTERMANUFACTURING SITETREATMENT CENTERCOURIERCOURIER1LEUKAPHERESIScollection2TRANSITfresh or cryo3ACTIVATIONTRANSDUCTIONLV or RV vectorEXPANSIONFORMULATIONharvest, fillCRYOPRESERVE4QC RELEASErelease testingTRANSITcryo shipper5INFUSIONafterlymphodepletionCHAIN OF IDENTITY: ONE PATIENT, ONE BATCHVEIN-TO-VEIN TIME
Fig. 3An autologous CAR-T patient journey. The chain of identity runs from collection to infusion. Numbered markers show the decisions Provenesis makes along the way. Durations are not to scale.
  1. Collection. Date and apheresis site for each patient.
  2. Courier. Lane and pickup window for the apheresis material.
  3. Manufacturing. Site, suite, and slot for each batch.
  4. QC release. Testing workload, analyst capacity, and the release date.
  5. Infusion. Shipment and infusion date, timed with lymphodepletion.

03Optimization

The best plan, not just a feasible one.

Provenesis formulates the plan as a mixed-integer program. Discrete choices, such as which slot a patient takes or whether to open a shift, are part of the model, not rules applied afterward.

Heuristic planners stop at the first plan that fits. Provenesis searches for the plan that meets your commitments at the lowest total cost, and shows you what it traded off to get there.

Slots and suites
Patients matched to slots across sites, within process durations and suite availability.
Staff and shifts
Operator and QC analyst capacity by shift, planned with the batches that need it.
Materials
Vector lots and critical materials allocated to the patients and campaigns that need them.
Commitments
Dates already promised to treatment centers are treated as commitments in the plan.
Slot schedule across four suites, with a binding QC release windowA six-week schedule. Four suite rows hold patient batches labeled P-01 to P-20, one bar per patient. Suite 3 has one open slot in week 3, marked as left open because QC binds. Below the suites, an operators lane shows shift staffing used against what is rostered, and a QC release lane shows testing workload against capacity. In weeks 3 and 4 the QC workload reaches capacity; that window is highlighted as binding, with a dollar value per added analyst shift.WEEK 1WEEK 2WEEK 3WEEK 4WEEK 5WEEK 6SUITE 1P-01P-05P-09P-13P-17SUITE 2P-02P-06P-10P-14P-18SUITE 3P-03P-07open: QC bindsP-15P-19SUITE 4P-04P-08P-11P-16P-20OPERATORSby shiftQC RELEASEtesting loadcapacityBINDINGλ $ per analyst shift
Fig. 4A slot schedule across four suites. Each bar is one patient's batch. The plan fills slots where operators and QC capacity allow, and shows where shared release capacity binds.

04Uncertainty

Plan for the patients who are not average.

Turnaround time, expansion yield, and the chance that a batch fails release vary from patient to patient. A plan built on the averages looks fine on paper and misses dates in practice.

Provenesis models that variability explicitly instead of collapsing it to a single number, so the plan is tested against the range of what can happen, not just the typical run.

Turnaround
Manufacturing and QC durations as ranges, not fixed lead times.
Yield
Variable expansion yields and what they mean for reaching the target dose.
Failures
Manufacturing failures and QC retests planned for, not handled as surprises.
Transit
Courier transit variability against the stability window of apheresis material.
Planning on the average versus planning on the rangeA histogram of turnaround time from collection to release, rising to a peak and then trailing off in a long right tail. A dashed line marks the average, where a fixed lead time would set the release date; the runs to its right finish later than that plan. A solid line further right marks a date that can be promised with confidence, and a bracket between the two lines is labeled as the cost to protect the date, in dollars.RUNSTURNAROUND, COLLECTION TO RELEASEAVERAGEfixed lead timeplans herelate againstthe average planPROMISED DATEholds for most runs,priced and chosenCOST TO PROTECT THE DATE, $
Fig. 5Planning on the average versus planning on the range. A fixed lead time puts the release date at the typical run. The range shows how many runs finish later, and where a date can be promised with confidence.

05Outputs

Decisions with the dollars behind them.

Every plan is a set of decisions you can act on: collect this patient at this site on this day, manufacture in this suite, release by this date. Each one comes with the trade-offs behind it.

For every binding constraint, Provenesis reports what relaxing it is worth: another slot, another shift, another QC analyst. Capacity requests come with a number attached.

Patient schedules
Collection, manufacturing, release, and infusion dates for each patient.
Suite and staff plans
Slot assignments, shift plans, and QC workload by day.
Trade-offs
What the plan gave up and why, stated in dollars.
Value of relaxing constraints
The dollar value of one more slot, shift, analyst, or courier pickup.
Value of relaxing each constraint (illustrative)Horizontal bars without values, one per constraint, sorted from most to least valuable to relax: QC analyst shift, marked as the most valuable, then manufacturing slot, operator shift, and vector lot. Courier pickup has no bar because that constraint is not binding, so relaxing it is worth nothing.CONSTRAINTVALUE OF RELAXING, $ PER ADDED UNITQC ANALYST SHIFTMOST VALUABLEMANUFACTURING SLOTOPERATOR SHIFTVECTOR LOTCOURIER PICKUPnot binding: no value from relaxing it
Fig. 6The value of relaxing each binding constraint, in dollars per added unit. Illustrative: bar lengths carry no data.

06The engine

Built on Mizahn.

Provenesis runs on Mizahn. Mizahn is a planning platform built specifically for process industries to help organizations make better decisions about their assets, denominated in dollars, under uncertainty so plans survive contact with the real world.

Its resource-task network represents steps that consume and produce several materials at once, so multi-step processes, variable yields, and shared capacity are modeled directly instead of worked around.

Next step

See Provenesis on a network like yours.

Book a working session with the founder. Bring your hardest planning week and we will show how Provenesis would plan it.

Book a demoGet the CAR-T planning brief