BrainBox AI

The platform connects via an edge device or direct API to a building's existing BMS, pulling data every five minutes without replacing field equipment. It supports protocols including BACnet and Modbus. This means deployment typically requires no mechanical retrofit — only a network connection and a commissioning period during which the AI learns the building's thermal profile.

It operates as a pure software subscription with a pay-for-performance option. Customers either pay a recurring SaaS fee or agree to share a percentage of verified energy savings. The company employs fewer building engineers than data scientists, and its contracts explicitly exclude ongoing on-site labor.

The platform is not designed for single-family residential or light commercial buildings where HVAC loads are intermittent and savings would not justify the monitoring overhead. It also does not target new construction — the product requires existing building telemetry for its training phase and is marketed to owners of occupied, operating buildings.

The firm uses IPMVP Option C (whole-facility utility-bill analysis) as the primary verification method, supplemented by sub-meter data where available. A baseline energy model is established during the commissioning period; subsequent savings are calculated against that model, adjusted for weather and occupancy, and reported through the customer dashboard.

ABB invested in the firm's Series A and subsequently integrated BrainBox AI into its own building-automation offerings, giving the startup distribution through ABB's existing channel partnerships. The Quebec government's Investissement Québec also participated, anchoring the company's engineering base in Montreal.