Lagrange by OrangeCat

Lagrange by OrangeCat is an AI tool built for the hydrocarbon sector. It performs root-cause diagnosis of cementing and well-integrity failures — problems like gas migration, poor mud displacement, slurry rheology issues, and microannulus formation. The tool draws on an agentic RAG system indexed across SPE papers, CBL/VDL log studies, and field histories. It doesn't just summarize; it reasons about pressure stability, slurry failure, and wellbore integrity using domain-specific logic. Every recommendation is source-traceable, so engineers can verify the evidence behind each diagnosis.

The system works by decomposing a problem into focused sub-queries, then iteratively retrieving relevant literature and field data. It uses ReAct reasoning — a pattern where the AI reasons and acts across specialized agents — to weigh multiple possible remedies. When several solutions exist, a confidence-weighted mechanism evaluates feasibility against equipment constraints and budget. The tool also handles uncertainty explicitly, acknowledging gaps in knowledge rather than pretending to have all answers. It runs on private national infrastructure with zero data exfiltration, making it suitable for operators concerned about data sovereignty.

Lagrange by OrangeCat is designed for petroleum engineers, upstream operators, and government energy bodies. It's especially useful for teams dealing with persistent cementing failures that resist conventional remediation. The tool has already been tested in a field case with ONGC, where it identified that the root cause was improperly spaced centralizers — not the slurry formulation. Beyond cementing, the company plans to expand into exploration, drilling, petrophysics, reservoir simulation, and production, all built on a purpose-built petroleum LLM with physics-informed constraints.