On this page
Use Cases & Target Customers
Who Navokoj Is For#
Navokoj solves discrete commitment allocation under hard constraints — where partial decisions must be preserved, and the cost of mistakes is high.
A note on our “Example Result” callouts: Until we have published case studies with named customers, every result on this page is a modeled outcome from internal benchmarks, not a customer testimonial. We’ve written them this way intentionally: we don’t want to claim a Fortune 500 customer quote that we don’t have. When we ship our first paid pilot (target Q3 2026), we’ll update this page with the real numbers and the real attribution.
Primary Verticals#
1. Logistics & Transportation#
Problems Solved:
- Vehicle routing with time windows
- Warehouse bin packing
- Fleet assignment
- Delivery slot optimization
Decision Variables:
- Each vehicle chooses one of a few legal routes
- Each package chooses one of a few eligible delivery windows
Why It Fits:
- Small legal choice set per decision
- Locked relationships matter (existing routes, driver preferences)
- Local repair needed when disruptions occur
Why They Pay:
- Fuel costs are directly measurable
- Customer SLA improvements are trackable
- Driver hours affect compliance
Example Result:
Modeled outcome on a 5,000-vehicle routing benchmark: 45-second baseline optimization reduced to 347ms with Navokoj’s
proengine, enabling real-time re-routing during disruption events rather than nightly batch runs.
2. Healthcare Workforce Scheduling#
Problems Solved:
- Nurse rostering with skill mix requirements
- Doctor shift scheduling with legal constraints
- Operating room allocation
- Equipment booking
Decision Variables:
- Each nurse/doctor chooses shifts from a legal set
- Each patient procedure chooses an OR slot
Why It Fits:
- Small legal shift set per person (5-7 days)
- Contractual constraints are hard (rest periods, max hours)
- Fairness constraints require explainable outputs
Why They Pay:
- Agency nurse costs are $100-200/hr
- Compliance violations carry legal risk
- Staff retention correlates with scheduling fairness
Example Result:
Modeled outcome on a 200-nurse, 14-day roster benchmark: 40% reduction in unscheduled shift swaps (a proxy for agency-nurse dependency) at solver runtime under 5 seconds per repair. Internal simulation; production customer data pending.
3. Cloud Infrastructure & DevOps#
Problems Solved:
- VM placement across physical hosts
- Container orchestration with affinity rules
- Network path optimization
- GPU cluster scheduling
Decision Variables:
- Each VM chooses one of a few eligible hosts
- Each container chooses one of a few legal pods
Why It Fits:
- Co-location, anti-affinity, and resource constraints
- Hardware heterogeneity matters
- Live migration constraints must be respected
Why They Pay:
- GPU idle time is pure cost ($3-4/hr per H100)
- 30% higher density = millions in savings
- Cloud margin improvement is C-suite visible
Example Result:
Modeled outcome on a 15,000-VM placement benchmark (Spectral-Multiplicative framework + Navokoj
proengine): 100% of affinity, anti-affinity, and resource constraints satisfied in 10.8 seconds. Projected savings: ~$1.4M/year for a mid-size cloud operator at industry-typical GPU idle-cost ratios.
4. Financial Services#
Problems Solved:
- Portfolio rebalancing with constraints
- Trade execution sequencing
- Credit limit allocation
- Insurance underwriting rules
Decision Variables:
- Each asset chooses rebalancing targets
- Each trade chooses execution timing
Why It Fits:
- Regulatory constraints are hard
- Risk limits must be strictly enforced
- Explainability required for audits
Why They Pay:
- Regulatory fines are massive
- Best-execution is legally mandated
- Risk management is existential
5. Telecom & Spectrum#
Problems Solved:
- Spectrum auction optimization
- Network frequency allocation
- Cell tower placement
- Signal interference constraints
Decision Variables:
- Each bidder chooses one of a few legal packages
- Each frequency chooses one of a few channels
Why It Fits:
- Billions of dollars in play
- Interference constraints are physical
- Synergies between adjacent bands matter
Why They Pay:
- Revenue optimization is directly measurable
- Regulatory constraints are non-negotiable
- Speed enables dynamic pricing
Example Result:
Modeled outcome on a synthetic FCC-style spectrum auction benchmark: $2.4B revenue achieved vs $1.18B greedy baseline (102% improvement) in 1.657ms solver runtime. The benchmark is synthetic; production auction performance depends on regulatory structure and bidder dynamics specific to each market.
6. Manufacturing & Supply Chain#
Problems Solved:
- Production scheduling with changeover costs
- Inventory allocation across warehouses
- Order promising with ATP/CTP
- Supplier selection with constraints
Decision Variables:
- Each order chooses one of a few fulfillment sources
- Each production run chooses one of a few schedules
Why It Fits:
- Capacity constraints are hard
- Customer commitments are already locked
- Local repair needed for disruptions
Why They Pay:
- Working capital tied up in inventory
- Expediting costs are massive
- Fill rate directly affects revenue
Problem Categories#
Beyond specific verticals, Navokoj excels at:
| Category | Examples |
|---|---|
| Assignment Problems | Task-worker, asset-location, resource-time |
| Scheduling | Shifts, production runs, maintenance windows |
| Configuration | Quote construction, plan selection, bundle rules |
| Routing | Vehicle, network flow, signal path |
| Placement | VM- host, facility-location, inventory positioning |
| Matching | Marketplace, auction, exchange |
Not a Fit For#
Navokoj is not the right tool when:
- Problems are continuous (use gradient-based optimization)
- Real-time control systems require sub-millisecond deterministic response (embedded systems)
- All constraints are soft preferences (multi-objective evolutionary algorithms may suffice)
- Problems have no discrete structure (pure simulation/Monte Carlo)
See Also#
- Marketing: Why Navokoj — product overview
- Pricing — tier selection by use case
- Results — benchmarks by problem type
- Limitations — where Navokoj is not a fit
- Glossary — engine selection guide