The scheduling engine
your product is missing.
Better schedules, in less time, than any other engine. The benchmarks are public.
Embeddable scheduling optimization for ERP, MES, APS, and workforce platforms. Ship world-class scheduling without building a solver team.
Powering scheduling for software vendors across
Why ISVs embed OptalCP.
Contract in weeks, not quarters
OEM agreements with large solver vendors run through enterprise legal, procurement, and audit clauses that can take two or three quarters to close. At OptalCP you negotiate directly with the founders, who hold both the technical and commercial authority to agree terms. Most partnership agreements close in weeks.
Go live in weeks
OptalCP ships as a native library that runs inside your process and returns results through your code. The modeling language reads like a scheduling textbook, and most integrations reach production in weeks. If you need a binding we do not ship yet, we build it.
Schedules your customers can trust
Real plants run on exceptions: frozen orders, operator qualifications, tank cleaning windows, split batches. OptalCP models those side constraints natively, so the schedule your customers see is one they can actually run instead of one their planners override by hand.
Direct line to the solver team
When you hit an edge case or need a custom propagator, you talk to the engineer who wrote it. Priority support with response times measured in hours.
Scales past the point others stop
OptalCP solves instances with over 100,000 tasks. Your largest customers stay on the same engine as your smallest, so you maintain one integration instead of a fast path and a slow path.
Why constraint programming wins at scheduling.
Most scheduling software uses a technology designed for a different problem.
Load jobs one at a time by priority score, never reconsider. Produces poor schedules on anything but the most granular problems, and changing the rule changes the answer.
Improve a candidate schedule by local moves until time runs out. Plateaus at a local optimum that can sit far below what OptalCP finds.
Encode scheduling into linear algebra. Time-indexed models need a variable per task per time slot; disjunctive ones need big-M constraints that weaken the bound. Excellent solvers, wrong encoding. Intractable at industrial scale.
Intervals are first-class objects; resource constraints reason over them with edge-finding and energetic propagation. Compact models, strong propagation, and optimum solutions at scale.
Where OptalCP is deployed.
Scheduling problems where the constraints are hard, the resources are contended, and a better schedule shows up directly on the customer’s balance sheet.
- Batch scheduling with time-link constraints
- Reticle and tool qualification management
- Cycle time and throughput trade-offs
- Finite-capacity production scheduling
- Sequence-dependent changeovers
- Alternative routings and machine selection
- Campaign scheduling and product-family sequencing
- Cleaning validation and changeover windows
- Shelf-life limits and tank capacity
- Resource-constrained project scheduling
- Multi-mode crew and equipment trade-offs
- Time-cost Pareto analysis across scenarios
- Shift rostering with skills and certifications
- Labor rules, fairness, and overtime limits
- Technician dispatch with travel windows
- Dock and terminal slot scheduling
- Crew rostering under rest-time regulations
- Fleet assignment with maintenance windows
The fastest constraint programming engine for scheduling.
OptalCP beats IBM CP Optimizer — the long-standing industry reference — on every standard scheduling benchmark family, proving more instances optimal at every time budget. Every result is reproducible from published source code.
| Benchmark family | Better solution when results differ |
|---|---|
| Job Shop97 | 95% |
| Job Shop + Transitions196 | 93% |
| RCPSP259 | 87% |
| Flexible Job Shop45 | 78% |
| Blocking Job Shop218 | 62% |
Results from the public benchmark suite, run on identical hardware with four workers. Live comparison pages chart objective progress, bound quality, branch counts, and LNS steps for every family. Source and instance data: optalcp-benchmarks · instance library
Your app calls OptalCP as a library, gets a schedule back, and displays it however you want.
Your application sends jobs, resources, and constraints via a native API. The model reads like a scheduling formulation. OptalCP finds optimal or near-optimal schedules and returns start times, resource assignments, and objective values. Display them in your own UI.
tasks = [model.interval_var(name=j, size=proc[j]) for j in jobs]
route = [[model.interval_var(optional=True) for m in machines] for j in jobs]
for j in jobs:
model.add(model.alternative(tasks[j], route[j]))
for m in machines:
model.add(model.no_overlap(seqs[m], setup_times[m]))
model.minimize(model.max([model.end_of(t) for t in tasks]))
What your models can express
The same constraint catalog as CP Optimizer. Existing modeling knowledge transfers directly.
Machine & resource sequencing
No-overlap constraints for single-capacity resources: machines, vehicles, rooms.
Alternative recipes & routings
Route tasks through one of several optional paths: flexible job shops, alternative workers or machines.
Sequence-dependent setups
Transition-time matrices for changeover, cleaning, and travel between task types.
Cumulative resources
Pulse-based capacity for manpower, tanks, energy, or any renewable resource with variable limits.
Calendars & time-varying availability
Step functions for shifts, breaks, weekends, and time-dependent cost or efficiency curves.
Custom objectives
Minimize makespan, weighted tardiness, total flow time, setup cost, or any arithmetic expression.
Reservoir & state constraints
Resource production and consumption over time: tanks, inventories, energy storage.
Proven optimality
Provably optimal solutions with a certificate of the remaining gap to the theoretical optimum.
Built by researchers who have spent their careers on scheduling.
OptalCP is built by constraint programming engineers with decades of combined experience in scheduling theory and industrial optimization. When you email us, you talk to the people who wrote the solver.
Diego Olivier Fernandez Pons
Former IBM ILOG. Originated the project and drives its technical and commercial direction. Background in operations research.
Petr Vilím, PhD
Former IBM CP Optimizer engineer. Architect of the OptalCP solver. Published researcher in constraint propagation for scheduling.
Nicolas Bonifas, PhD
Former IBM CP Optimizer engineer and serial AI founder. Leads partnerships and commercial growth. PhD in constraint programming for scheduling.
Embed OptalCP in your product.
From first call to production: here’s what the partnership looks like.
Tell us what you're building
A 30-minute call to understand your product, your scheduling problem, and what your customers need.
Proof of concept
We model your benchmark problem in OptalCP and show you schedule quality and solve times on your data.
OEM license
Flexible licensing: per-deployment, per-core, or annual. Structured for ISVs who ship to many customers. Includes ongoing solver updates, priority support, and integration guidance.
Response within one business day from the engineers who wrote the solver.
Preview
Full modeling API, get objective value. Solution values require a license.
- All modeling primitives
- All benchmark source code
- Community support
Academic
Full solution values for coursework, research, and publications.
- Everything in Preview
- Unrestricted solution values
- Email support from the team
- Cite-and-use license
Commercial & OEM
Embed OptalCP in your product. OEM licensing designed for software companies.
- Everything in Academic
- OEM redistribution rights
- Priority support
- Custom licensing models
- Integration guidance
Ready to evaluate?
Your engineering team can have a working model running in five minutes.