CASE NO. 01 · ENERGY / INDUSTRIAL IOT · CLIENT: CONFIDENTIAL · TIMELINE: 12 WEEKS TO MVP

Bringing the $50B oil patch online, without the internet.

Independent oil producers run thousands of wells scattered across the remote American West. Most of them have no cell signal. We built a field platform that works anyway: an offline-first mobile app paired with custom LoRaWAN sensors that keep watch on every tank, around the clock.

What we delivered

Product roadmapFractional CTOIndustrial designCustom PCB & firmware (C/C++)LoRaWAN meshReact Native (offline-first)Python / DjangoAWS IoT CoreAnomaly detectionTimescaleDB

The shape of it

From prototype to production scale. The field kept working whether or not the network did.

The client

A fast-growing SaaS provider serving the “long tail” of oil and gas: independent operators with thousands of low-volume wells and none of the IT budget of a major.

The friction

Pumpers drove hundreds of miles a day to hand-gauge tanks and write the numbers on grease-stained paper tickets. Data took weeks to reach the office. Between visits nobody was watching at all, so leaks, theft, and equipment failures went undetected for days.

The constraint

90% of these wells sit in cellular dead zones. Standard cloud-connected IoT does not work here.

  • No cellular connectivity: 90% of wells in dead zones
  • Harsh environmental conditions (Permian Basin)
  • Legacy hardware compatibility
  • Zero-touch compliance requirements

From prototype to production-grade.

01

Weeks 1–12

The offline MVP

A React Native app that ran as a complete local system, doing the oilfield math on the device itself: API gravity, temperature corrections, all of it before anything reached a server. Pumpers adopted it because it did the arithmetic they had been doing on paper. Onboarding took under eight minutes per user.

02

Months 4–12

The hardware bridge

The 'Silent Watchdog': a custom battery-powered LoRaWAN sensor. It wakes, pings its reading out through steel tank walls over long-range radio, then sleeps. No off-the-shelf sensor could take that duty cycle, so we engineered this one for a five-year battery life.

03

Scale

The shadow twin

A cloud layer that checks what the pumpers reported on run tickets against what the sensors actually saw. A fifty-barrel drop with no ticket logged? Flagged. Level falling faster than the max pump rate? Emergency interrupt.

Inside the build.

TANKLORA SENSORFIELD GATEWAYCLOUDPUMPER APPOFFLINE-FIRST SYNC
Fig. 01·Thin edge, fat cloud: critical safety logic lives on the edge; heavy analysis lives in the cloud.

When the device and the server disagree.

A pumper goes offline for three days. By the time the phone reconnects, its copy of the data and the server's have drifted apart. We built a Merkle-tree sync engine on SQLite and made the device the source of truth for field data. Every edit is append-only, so nothing is ever overwritten. Operators can replay a pumper's day and see exactly when a number changed.

DEVICE · SOURCE OF TRUTHAPPEND-ONLY LOGSERVERMERKLE DIFF
Fig. 02·Merkle-tree sync engine.

What changed.

The platform changed the economics of running these wells. Leaks surfaced in hours, not days. Pumpers drove less, because the app sent them to the tanks that actually needed a visit. And when operators and landowners had to reconcile payments, the system became the record they both worked from.

We stopped guessing what was happening in the field. This system turned the lights on.
CEO · Early-adopter operator

Have a constraint everyone else calls impossible?