Every injector, every producer, every pair. Ranked by who is actually supporting whom.
A mature CO₂ flood: dozens of injectors and producers, years of daily injection and production history, and the field plat with its faults. Cogzia turned the surveillance review that is normally done chart by chart into a ranked answer — which injectors are supporting which producers, how strongly, and how soon — with the evidence behind every call, and the operator's own judgment folded back into the result.
One flood. Dozens of wells. A review done by eye.
Flood surveillance is a reading job before it is a decision. For every producer, the injection history of the injectors around it has to be pulled, lined up, and judged by eye: which ones is the well responding to, and how long after? Across a whole pattern that is hundreds of chart pairs, redone every time the picture changes. The judgment belongs to the operator's engineers. The assembly is what eats the week.
- SCOPEThe surveillance workbook the operator already kept, and the field plat. No new data, no new system.
- SOFTWAREThe production system is a decade old. It stayed. Cogzia reads it where it sits and gives it a modern front end — daily data in, a ranked answer out — without asking anyone to change what they run.
- FORMATThe same review, returned as a ranked list per producer instead of a stack of charts.
- OWNERSHIPThe operator's engineers stayed in charge. Where they overruled a call, the ranking re-ran with their call in it.
The review, ranked. The evidence, attached.
Which injectors are actually supporting each well, in order, with how strongly and how soon the well responds.
- Faults are respected as barriers: no injector is credited with influence across one.
- The correlations are calculated the same way every time, on the same data. AI writes the brief; it does not invent the numbers.
- Every claim in the brief points to the wells and the dates behind it.
More feedback. Better adjustments. More production.
Flood surveillance is a feedback loop: adjust, wait for the response, learn, adjust again. On a flood measured every day, the response shows up in about a week — so the limit on how often you can adjust is how often you can afford to redo the review. A review that re-runs itself when new data arrives turns a monthly adjustment into a weekly one, and the difference compounds.
Days of chart reading became an afternoon of checking.
The team opened a ranked list, not a stack of charts. Where the ranking matched what they already believed, they moved on. Where it didn't, the evidence was one click away, and their override went back into the result. When the next week's history arrives, the review re-runs and the list updates — the team picks up from the changes, not from chart one.
Bring us an injector problem.
One pattern, one workbook, one question: which injectors are doing the work? We'll show you the ranked answer and the evidence behind it.