Sematryx turns constrained decision problems into a clear workflow: formulate, solve, explain. The engine orchestrates solvers and learning signals so you get strong results—not a single fixed optimizer.
Express objectives, bounds, and constraints—through code, API payloads, or conversational flows. The product focuses on making the problem well-posed before burn is spent on search.
The engine selects and coordinates strategies from a multi-solver portfolio, using adaptive selection and transfer from prior runs where applicable—see Performance for evidence.
Get decision-relevant explanations: what drove the optimum, sensitivity-style insight where configured, and audit-friendly narrative for teams that need to justify decisions—not just a raw vector.
The fastest way to put optimization in an agent workflow today is MCP: one tool call from Claude Desktop, Cline, or compatible clients. That is a distribution wedge, not the whole product—underneath is the same engine and APIs.
Industry-specific library surfaces are a future productization direction and are limited today—see honest status on Domains. The core offer is the engine and workflow above.