A top-3 US energy company put grounded, mobile-first AI in the hands of thousands of field technicians — bringing the right answer to the worker at the right moment, with the safety bar field work demands.
A top-3 US energy company employs thousands of field technicians servicing complex equipment across difficult terrain, often with limited connectivity. Technical knowledge — equipment manuals, safety procedures, repair playbooks, regulatory protocols — was scattered across decades of documentation and the heads of senior technicians approaching retirement. The company needed AI that worked on a phone, in the field, often offline, in seconds — bringing the right answer to the technician at the right moment, with the safety bar field work demands.
BizzSoftware partnered with Field Operations, Technical Training, and Safety to design and build a mobile-first AI copilot deployed across the technician workforce. The build was constrained by the realities of field work:
Connectivity is intermittent at best. Critical content cached locally, model calls degraded gracefully.
Every recommendation traceable to source. Procedures requiring formal sign-off explicitly flagged.
Built around the actual job: diagnose, look up procedure, document, sign off.
Natural-language access to manuals, procedures, and safety guidance, with grounded answers and citations on every output.
Technicians describe symptoms, get likely-cause hypotheses tied to verified diagnostic procedures.
When senior technicians solve novel cases, the system captures the resolution as candidate knowledge for review and integration.
A technician 40 minutes from the nearest cell tower opens a transformer cabinet she's never serviced. The connection drops to one bar and back to none.
She opens the application on her phone. The unit's manual, the OEM safety procedure, and the regional regulatory protocol are already cached on the device — the system pre-fetched them from the work order before she left dispatch. She queries: "this unit is showing differential current spike on phase B; safe to proceed?"
The application answers in seconds with a procedure citation, the specific torque and PPE requirements pulled from page 47 of the manufacturer manual, and a flag: this is a Class-II procedure requiring a senior-tech sign-off before energizing. She follows the procedure, photographs each step, signs off in the application. The signoff sync is queued; when she's back in coverage 90 minutes later, it lands in the safety record system automatically.
No phone-a-friend. No guess. No paper-record gap. The senior tech reviewing the case file the next morning sees a clean lineage of every decision, citation, and signature.