Optimizing Configuration with AI

SYNOPSIS

A manufacturer of high-speed measurement equipment offered a broad, capable product catalog, but customers had no reliable way to translate their needs into a valid system configuration, since each setup depends on real compatibility and physical constraints between components. BP3 built Solution Designer, a configuration engine using IBM's ILOG CPLEX Constraint Programming library and Operational Decision Manager, which models those constraints directly and dynamically narrows options as a customer configures their system. The result is a faster, more confident configuration process that guarantees every suggested build is one that actually works.

Configuring Complex Hardware with Constraint-Based AI

A broad product catalog is usually a sign of a business doing well: more products, more use cases served, more customers with a system that fits their specific need. It also creates a quiet problem at the point of purchase. The wider the catalog, the harder it becomes for a customer to work out, unaided, which combination of components will actually function together.

A leading provider of high-speed measurement equipment ran into exactly this. Its systems are built from three core components: a chassis, one or more controllers that act as the system's processing core, and module boards that handle input and output. Each of those components comes in multiple variants, and not every combination is valid. Certain controllers only work with certain module boards. Certain module boards only fit in certain chassis slots. Physical space inside the chassis is finite, and a configuration that looks correct on paper can still be physically impossible to build. None of this is visible to a customer browsing a catalog page, which left people either guessing, second-guessing an order, or reaching out for help on something the tool itself should have been able to tell them.

The client engaged BP3 to solve this at the source: not by simplifying the catalog, but by building a tool that understood the engineering constraints well enough to guide customers through them automatically. The result was Solution Designer, a configuration engine built on IBM's ILOG CPLEX Constraint Programming library paired with IBM's Operational Decision Manager.

Constraint programming is a good fit for this kind of problem because it does not just search for one valid answer, it models every rule governing how components relate (compatibility, dependency, physical fit) and uses that model to filter the option set in real time. As a customer selects a chassis, the engine already knows which controllers are compatible with it. As they add a controller, the available module boards narrow to only the ones that will actually work in that combination and fit in the remaining physical space. Nothing invalid is ever presented as an option, which means customers cannot accidentally configure a system that will not work.

The effect for the customer is a process that feels closer to working with a knowledgeable systems engineer than to filtering a catalog. Instead of parsing datasheets and cross-referencing compatibility manually, the customer answers a series of guided choices, each one already filtered by everything they have selected so far, and arrives at a configuration that is guaranteed valid, both logically and physically.

The broader principle extends well beyond this client's product line. Any business selling configurable, constraint-heavy products, industrial equipment, technical hardware, complex service bundles, faces the same underlying challenge: the richness of the catalog and the ease of configuring it are usually in tension. Constraint-based configuration engines resolve that tension directly, by encoding the real rules of the product once and letting the software handle the complexity a customer should never have to see.