Across downstream and midstream.
Same asset DNA. Same sales motion. Adjacent customer base.
Four tier-one segments, one product.
Refineries
Unplanned downtime runs $1M+ per day. Fleet-wide asset intelligence flips that economics.
- Centrifugal compressors
- Heat exchangers
- Air coolers
- Pumps
Petrochemical
Continuous processes where a single unplanned stop cascades through downstream units.
- Crackers
- Reactors
- Distillation towers
- Compressors
Gas Processing & Fractionators
Uptime is throughput. Early degradation flags protect contracted volumes.
- Compressors
- Chillers
- Fractionation columns
- Dehydrators
LNG
Downtime economics at $3–5M per day per train. Every prediction pays for itself.
- Main cryogenic exchangers
- Refrigerant compressors
- BOG compressors
- Pumps
No rip-and-replace.
KeplrAI plugs into the plant's existing control and historian stack. Read-only at first. Write-back only where the plant authorizes it.
Real assets. Real signals. Real tags.
Four use-case models shipping today. Each repeats across all four segments because the asset DNA is the same.
Vibration signatures learned per bearing. Surge-margin erosion flagged 6–12 weeks before trip.
Fouling signature learned from thermal and pressure-drop trends. Cleaning windows proposed before approach drops.
Per-bundle fouling detected from fan load and air-side ΔT asymmetry. Targeted cleaning, not blanket.
Network-level degradation detected from pump interaction patterns. Impeller wear flagged before performance collapse.
Upstream production: declared, not active.
Onshore central processing facilities, offshore platforms, FPSOs. Same asset physics, different connectivity. Sequential expansion as fleet models mature.