Subsurface Characterization
Estimate rock properties from logs, predict well performance before drilling, and search decades of well files.
Models estimate shale volume, porosity, water saturation and permeability from logs, and predict productivity at planned wells from geology and offsets. A well file assistant searches core reports, completion reports and studies, and answers questions with the page and well behind each answer.
Liquid Loading and Artificial Lift
Spot gas wells starting to load up, and match each well to the lift method most likely to work.
Models forecast flow rate and casing pressure per well and flag loading early. Classifiers recommend gas lift, plunger, rod pump or velocity string. Writes the case for each intervention: what the data shows, similar wells and how they responded, and the expected uplift from the model.
Refinery Quality Soft Sensors
Predict product quality in real time instead of waiting hours for the lab, and stop over-refining to be safe.
Soft sensors predict quality, such as diesel sulfur, from process variables in real time, and show which flows and temperatures are driving it. Drafts shift handover notes from process data, alarms and logbooks, and answers operator questions from operating procedures.
Flaring and Emissions
Predict the pressure spikes that force flaring, estimate flows where there are no meters, and draft the reports.
Virtual flow meters estimate flows from surrounding sensors. Models predict pressure spikes hours ahead so operators can rebalance before flaring. Drafts flaring event explanations and regulatory emissions reports from the data, with every figure traced to its source.
Permit Risk and Incident Learning
Flag high-risk work before it starts, and put the lessons from similar past jobs in front of the crew.
Models classify activity risk from the job description, site and conditions, so supervisors focus reviews on the riskiest work. Reviews permits and job safety analyses for missing hazards and controls, and finds similar past incidents with what went wrong.
Supply Chain and Spare Parts
Predict which orders will arrive late, right-size spare parts inventory, and take the follow-up work off buyers.
Models predict on-time delivery for each order from supplier, item, quantity and dates, and forecast spare parts demand by site. Drafts expediting emails for at-risk orders, reads supplier replies and updates the order, and pulls key terms from contracts.
Engineering Copilot
Give engineers and technicians answers from procedures, standards and manuals, with the page behind every answer.
A copilot answers troubleshooting and procedure questions in the field, drafts management-of-change documents, and cites the source page. Answers come only from approved, current documents and say “I don’t know” when they are silent. Safety-critical steps always point to the procedure.
