Real-time classification, per well.
Physics-informed AI classifies developing risk in real time with the channel evidence behind each call, alongside per-section ROP optimization that maximizes footage without driving dysfunction.
The events it classifies.
Developing downhole risk, classified from the multi-channel pattern every 15 seconds, mechanism-specific, so the response matches the cause.
Circulation losses
Loss severity tracked as the trend develops, so it is graded before it is total.
Stuck pipe
The mechanism is named, not just the symptom, so the crew acts on the right cause.
Equipment performance
Wear and degradation trends surfaced early, so a tool change happens on schedule, not mid-run.
From the call to the next action.
An alarm tells the crew something is wrong. DO NOW tells them what to do about it: a specific next action, written against your own approved procedure for that mechanism, not generic advice.
The call
The mechanism is named and the contributing channels are shown, so the crew can see why the call was made.
Do now
One short, specific action for the mechanism that was classified, not a generic alarm to acknowledge and move past.
Your SOP
The action text is mapped to your approved response for that hazard, so it matches how your crews are already instructed.
Real-time ROP optimization.
Setpoint guidance that optimizes ROP without driving dysfunction: the three levers the driller controls, tuned live.
Flow
Enough flow to clean the annulus and lift cuttings, without pushing ECD past the fracture window.
WOB
Maximise footage without inducing stick-slip, bit balling, or passing the point of diminishing ROP.
RPM
Manage vibration and steer ROP with the drive, balanced against torque and the downhole tool limits.
Escalation, governed.
Detection is logged immediately and permanently, regardless of what happens next. Delivery is a separate decision, made per recipient against the alert's live state at each user's configured hold window. An alert that resolves before the hold expires is suppressed, and the suppression is logged too. If no acknowledgement arrives in time, escalation overrides personal preferences and fires regardless.
Every shift starts on the same picture.
The night shift usually inherits an alarm history with no context. The instructions your engineers issued sit next to the live call instead, so the on-coming crew works to the same plan the last shift did.
Standing
The instructions that hold for the whole section, pinned where the crew sees them, not buried in the program.
Shift
What was issued to the rig for this shift, shown against the window it actually applies to.
Handover
What was classified last shift, what was acknowledged and what is still open, so nothing is lost at the tour change.
In this surface.
- Mechanism-specific risk classification, every 15 seconds
- 70+ ingested rig-telemetry channels
- “Why this call” evidence per event
- DO NOW actions mapped to your SOPs
- Standing & shift instructions, carried to handover
- ROP optimization & limiter detection
- Setpoints & recommendations
- Pore pressure / FG context
- Tally, trips, BHA & trajectory
- Replay & audit
- Alert routing & FP control
Evaluate on operator data.
Provide one live section; Vogter reports what it would have classified in real time, scored against the operator's drilling reports.