Solvicus

Blog

Guides & worked examples

Notes on mathematical optimization, linear and mixed-integer programming, and using MCP servers to give AI agents provably optimal answers.

July 30, 2026

Committed Use Discount Planning: Forecast-Driven vs Usage-Based

Usage-based CUD buying sizes commitments from your past and hedges the future. Forecast-driven planning optimizes for the demand you already know — and when your demand is contractual, that's not a guess.

July 16, 2026

How Much CUD Should You Buy?

Commit too much and you pay for idle capacity; commit too little and you overpay on-demand. Coverage, utilization, and break-even are how you find the line between them.

July 14, 2026

1-Year vs 3-Year Committed Use Discounts

A 3-year commitment is cheaper per unit than a 1-year one — but it's a bigger bet on demand you can't yet see. How to think about the term trade-off, and where break-even comes in.

July 6, 2026

Why AI Agents Need Deterministic Solvers

Ask a large language model the same question twice and you can get two different answers. That's fine for conversation, and unacceptable for scheduling and cost decisions — here's why.

July 5, 2026

Home Energy Management Optimization: A Worked Example

An EV charging deadline, a home battery, dynamic hourly tariffs, and a heat pump's thermal inertia — walked through as the optimization problem behind Solvicus's home-energy-management use case.

July 4, 2026

Crop Rotation Scheduling: A Worked Example

24 fields, a 3-year Fabaceae break rule, and a 5-year price forecast — walked through as the mixed-integer program behind Solvicus's crop-rotation use case.

July 3, 2026

GCP Committed Use Discount Optimization: A Worked Example

250 customer contracts, three e2 machine sizes, and a CUD renewal decision — walked through as the mixed-integer program Solvicus actually solves to pick your commitments.

July 1, 2026

Staff Scheduling Optimization: A Worked Example

A 15-physician practice, room constraints, and labor-law rest periods — walked through as a mixed-integer program to show why spreadsheets and manual rostering break down.

June 19, 2026

Linear Programming vs. Mixed-Integer Programming

Continuous variables get you linear programming. Add yes/no decisions and whole-unit counts, and you need mixed-integer programming instead. Here's the difference.

June 6, 2026

What Is Mathematical Optimization?

Variables, objectives, and constraints: the three ingredients behind every optimization problem, and why the answer they produce is provably optimal — not a best guess.