Who we are

Built by operators, not data scientists.

The gap between a working demo and a running plant is not a model problem. It is a knowledge and experience problem.

The knowledge that never transfers.

Every operation runs on two sets of rules: documented procedures and the operating knowledge your senior people carry but never transfer. The calls they make before the instruments confirm them. The limits that shape throughput in ways the permit does not explain.

AI sees the first set. It misses the second. We have seen this failure from the inside, across refining, chemicals, metals, and automotive. Closing that gap is the work. If you have seen it in your operation, we want to hear about it →

Decades running the industries AI is trying to automate.

30+
years of operations leadership across the founding team
2
granted US patents in vehicle and clean-energy process design
6
industry verticals: refining, metals, automotive, energy, finance, AI

Refining & Chemicals

Operating knowledge from turnarounds, control rooms, and PSM-covered processes. Which instruments to trust, where the SOP runs out, and how regulatory limits actually shape throughput.

Metals & Mining

End-to-end operations, process design through continuous production. The furnace conditions the design didn’t anticipate, and the QC calls made before the lab confirms them.

Automotive & Heavy Equipment

OEM and heavy equipment scale, concept through manufacturing release. Where programs break: NVH that surfaces at integration, and launch decisions that live between engineering release and what the line can actually build.

Clean Energy

Battery materials recovery, concept through commercial production. The gap between what works in the lab and what holds up at full scale. And the decisions that can’t be undone.

Finance

Capital markets and private equity across energy, manufacturing, aviation, and logistics. Where models get it wrong: throughput an operation can’t sustain, and risk that lives in the shift change, not the balance sheet.

AI & Data Science

PhD research in safety-critical processor architectures. What it takes to run AI reliably when someone is trying to break it: anomaly detection for ICS, audit trails that survive a breach, fault-tolerant compute.

The equation

Today’s best decision is the one that survives tomorrow’s changes.

V(s)=maxa[R(s, a)+γ·V(s′)]
V value · R reward · γ discount

We have run these operations.
We know where AI deployments break.

If you’re closing this gap in operations, we’d like to hear how.