Oil and gas · Strategy and operations
Krog is the strategy partner and the vendor — one firm, one mandate, across upstream, midstream, LNG and refining. We answer for the numbers after go-live, not only for the recommendation that produced them.
The mandate
Consultancies hand over a roadmap and leave. Integrators build whatever the roadmap told them to build. Krog signs for both, which has one useful consequence: we cannot recommend anything we are not prepared to operate.
Krog.AIAI strategy
What is worth doing, in what order, against the constraints the asset actually has — regulatory, subsurface, and the ones nobody wrote down.
AI operations
We are the vendor as well. We build it, commission it, and hold it to its numbers through crew rotations and budget cycles.
One accountable party — first diagnostic to year five
AI strategy
An AI strategy for a producing asset is not a technology roadmap. It is a series of decisions about which constrained processes deserve intervention, what a defensible case looks like when the field has twenty years left on it, and which risks the operation cannot reasonably be asked to carry.
We start in your historian, your well files and your maintenance history, then test every hypothesis against the people who actually run the asset. What you get back is a sequence you can fund, defend in a safety review, and hold us to.
AI operations
Krog is the vendor, not a recommender of vendors. Our engineers work at the cabinet and in the historian, inside your permit-to-work and management-of-change process, under the same access discipline as anyone else on site.
Go-live is where the obligation starts. Feedstock shifts, instruments drift, wells decline and units get re-rated. We keep the system inside its acceptance criteria for as long as we hold the mandate — and we are the ones paged when it moves outside them.
Where we work
Krog does not serve retail, banking or software. Every engineer we hire, every reference architecture we own and every failure library we maintain points at hydrocarbon production — where a bad recommendation costs a platform, a unit or a licence, not a quarter.
The specialisation is the product. It is why the first session starts roughly where a generalist firm arrives in month six, already fluent in your operating envelope, your regulator and your vocabulary.
Platforms, FPSOs, subsea tiebacks
Pads, artificial lift, field-wide optimisation
Campaign planning, integrity, intervention
Crude units, FCC, hydrotreating, blending
Liquefaction trains, treatment, regasification
Compression, leak detection, throughput
Tank farms, loading, hydrocarbon accounting
Corrosion, inspection, flaring, methane
Why this partner
The 2026 market has no shortage of capable generalists. Very few of them will be on the line at three in the morning when a model starts drifting on a night shift.
Oil and gas is the only industry we serve, so our reference architectures, equipment models and failure libraries already have the shape of your asset. Rotating equipment degradation, separation and treatment behaviour, well performance and integrity data are not new problems here.
You are not funding our education in your industry. Every hour we bill goes to your operation rather than to learning what a turnaround is.
Our customers get privileged access to patented AI innovations under development since 2019 — a level of accuracy and performance the broader market cannot currently put into production.
The people who recommend the work are the people who operate it. There is no vendor triangle to arbitrate, and no handover cliff where the business case quietly stops being anyone's problem.
Deterministic behaviour, degraded links, zone classification, change control and the safety case are design inputs from the first sketch. Segmentation and least privilege are assumed, not retrofitted the week before commissioning.
Operators plan in decades and fund in turnaround windows. Our engagements are multi-year operating agreements, not a stack of statements of work that expire long before the asset does.
The engagement
The first three are what a consultancy sells. The fourth is what a producing asset actually needs, and it is the one this firm is built around.
Weeks
We come to the asset. Historian and maintenance data, the control philosophy, the operating envelope, the constraints nobody documented, and the operators who already know where the losses are.
A quarter
The strategy: portfolio, sequence, target architecture, autonomy levels, governance, and a case your finance function is willing to hold us to.
Per release
Design, build, integrate and commission inside your management-of-change process, against acceptance criteria agreed before a line of code is written.
Open-ended
Run it, watch it, extend it. Wells decline, units get re-rated, instruments drift and models age. This is where a Krog partnership spends most of its life.
Phase four has no end date, and the strategy gets rewritten from inside it. In 2026 the only AI strategy worth paying for is one written by the people already carrying the pager.
Start here
A first conversation is an hour with your operations and technology leads. If oil and gas is not where your losses are, we will tell you and point you somewhere better.