Skip to content
BasinWright
Industry solutionsBook a session
Industries

Energy, oil & gas

Procurement intelligence. Technical specification search. Vendor evaluation. Asset maintenance — against a supply chain where a fortnight's slip costs more than the part.

The problem

The money leaks between the systems, not inside them

In an asset-heavy operator, the expensive mistakes are almost never made inside a single system. They happen in the gaps: a long-lead valve that clears inspection after the turnaround window closed, a critical spare sitting in a yard three hundred miles from the rig waiting on it, heat numbers on a mill certificate that do not match the spools that arrived, a call-off priced well above the frame agreement nobody checked it against.

Each of these is individually findable by a competent expeditor with enough time. None of them is findable across a portfolio, continuously, by anyone. And the cost of finding one late is not the price of the part — it is demurrage, liquidated damages, or a frozen scope that grows by fourteen work packs.

Most of the evidence is also unstructured: specifications, certificates, contracts, packing lists, inspection reports. That is a language problem feeding a decision problem, which is exactly the split the platform is built around — a model reads the document, a deterministic model makes the call.

Use cases

Where models earn their place

Procurement and supply chain is where this industry's catalogue is deepest, because it is where the leakage is largest and the evidence is worst organised.

Long-lead expediting

Turnaround

A critical item tracking late against a fixed shutdown window, flagged while the window can still be replanned.

  • Schedule slip
  • Turnaround cost
  • Expediting hours

Single-source exposure

Supply risk

One qualified vendor holding most of a critical spares list, quantified across the portfolio rather than felt anecdotally.

  • Concentration risk
  • Qualified alternates
  • Lead time

Stranded spares

Inventory

The part an asset is waiting on already sitting in another yard, under a different description in a different system.

  • Working capital
  • Downtime avoided
  • Redeployment rate

Certificate mismatch

QA/QC

Heat numbers on the mill certificate disagreeing with what was delivered — caught at the gate rather than at installation.

  • Rejection rate
  • Rework cost
  • Compliance findings

Demurrage risk

Marine logistics

Berth congestion pushing discharge past laytime, forecast far enough ahead to re-sequence.

  • Demurrage paid
  • Laytime utilisation
  • Vessel turnaround

Price against frame agreement

Procurement

Call-offs priced above the agreement they were supposed to be governed by, across thousands of lines nobody re-checks.

  • Spend recovered
  • Contract compliance
  • Maverick spend
Connected to

The systems the supply chain already runs on

Contracts, certificates and yard data are usually the hardest to reach and the most valuable to connect. All of it is read in place.

ERP — materials & maintenance
Asset management
Supplier & vendor portal
Expediting & inspection
Freight & marine logistics
Contract repository
Warehouse & yard
QA/QC & certification
Governance

Sovereignty is usually decided before the model is

Operators in this sector are frequently constrained by residency law, national-oil-company policy or plain security posture long before anyone has an opinion about a model. That is why the deployment options are not an afterthought here:

  • On your own GPU, on-premises — training and inference inside your perimeter, with no prompt or embedding data leaving the network
  • In-country residency where law or contract requires it
  • Air-gapped where the environment demands it

The governance is the same as everywhere else on the platform: credentials sealed with per-tenant keys, access enforced by attribute-based rules with deny-overrides, and every privileged action written to an immutable audit log. Sanctions and country-of-origin screening ride on top of that rather than beside it, which is what keeps a customs or trade-compliance question answerable from the same evidence trail as the commercial one.

Worked designs

How we would build it

Reference deployments: a problem shape, the design we would propose for it, and what we would expect to be measured afterwards.

All reference deployments
Fraud & Financial Crime

Fraud decisioning: when the alert is not the unit of work

Rules engine, two bolted-on scores, three review queues that each think they are the queue of record. The models are usually fine. What is missing is assembled evidence.

8 min readLodestar
Clinical Operations

Ambient clinical documentation, and the discipline of refusing

The transcripts are good. The notes are not. The failure is never dramatic hallucination — it is smoothing, a note that reads correctly and contains three lines nobody said.

7 min readCartograph
Sovereign AI

A sovereign estate with no external dependency

The requirement that separates sovereign AI from AI that happens to be hosted locally — and the one most of the market cannot meet, including on its own architecture diagrams.

8 min readSovereign
Talk to us

Start with one decision

Bring us a category of spend or a turnaround that slipped. The leakage is usually already visible in data you have.

We reply within one business day. No sales sequences.