Technology
The Science Behind kint
Mathematical solvers for exact answers. Machine learning for data-driven problems. LLMs for natural language input. kint selects the right approach automatically.
Methods
Optimization Methods
kint selects the right formulation for your problem automatically. Your team describes the problem. kint picks the method.
LP
Linear Programming
Continuous variables, linear constraints and objective. Allocation, blending, flow optimization. Solved in polynomial time.
MIP
Mixed-Integer Programming
Discrete and continuous variables. Scheduling, routing, facility location. Branch-and-bound with cutting planes.
CP
Constraint Programming
Logical and global constraints. Job-shop scheduling, timetabling, sequencing. Propagation and search.
QP
Quadratic Programming
Quadratic objective, linear constraints. Portfolio optimization, regression, curve fitting.
NLP
Nonlinear Programming
Nonlinear relationships. Process engineering, chemical blending, continuous control. Interior-point and SQP methods.
Solvers
Solvers
Commercial and open source. kint uses the best solver for each problem type. Your team doesn't need to choose.
Gurobi
CommercialLP, MIP, QP. Industry standard for large-scale optimization.
CPLEX
CommercialLP, MIP, QP. Strong performance on scheduling and planning problems.
OR-Tools
Open SourceCP, routing, assignment. Google's operations research suite.
GLPK
Open SourceLP, MIP. Lightweight solver for smaller problems.
CP-SAT
Open SourceConstraint satisfaction and optimization. Fast for scheduling and sequencing.
Machine Learning
Machine Learning
Surrogate Models
XGBoost, Bayesian optimization, ensemble methods. Learn the relationship between inputs and outputs from historical data. Replace unknown formulas with learned models.
Demand Forecasting
Time-series prediction as input to optimization. Forecast demand, then optimize supply chain decisions against that forecast.
Warm-Starting Solvers
ML predictions provide starting points for mathematical solvers. Faster convergence. Better solutions in less time.
LLMs
Large Language Models
Natural Language Input
Describe your problem in plain text. "Minimize delivery cost for 47 trucks across 6 countries with time windows." kint extracts the mathematical structure.
Automatic Extraction
Variables, constraints, and objectives identified from your description. Your team reviews and adjusts. The model reflects your business.
Not Used for Solving
LLMs translate. Solvers solve. The optimization result comes from mathematical solvers, not language models. Proven methods, not generated answers.
Architecture
Architecture
API-First
RESTful API with OpenAPI documentation. Every feature available programmatically. Your systems integrate directly.
OEM-Ready
White-label the optimization engine inside your product. Your brand, your customers. kint runs in the background.
Stateless Runs
Each optimization run is independent. No session state. Scale horizontally. Deploy anywhere.
WebSocket Progress
Live updates during optimization. Current objective value, optimality gap, iteration count. Your users see progress in real time.
See the Technology in Action
Bring your problem. We'll show you which solver kint selects, how it formulates the model, and what the result looks like. 30 minutes.