kint

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

Commercial

LP, MIP, QP. Industry standard for large-scale optimization.

CPLEX

Commercial

LP, MIP, QP. Strong performance on scheduling and planning problems.

OR-Tools

Open Source

CP, routing, assignment. Google's operations research suite.

GLPK

Open Source

LP, MIP. Lightweight solver for smaller problems.

CP-SAT

Open Source

Constraint 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.

Book a Demo