6 AI initiative profiles for Shell across 5 departments, with evidence grades and source links. Strongest evidence: A · Operator filing — T-Pulse AI CCTV monitoring of unsafe acts, conditions and leaks.
Company and vendor statements are attributed, not independently validated. Evidence grades: A operator filing, B operator release, C vendor-reported, D secondary or unverified.
Later research and source checks
Supplementary notes can update existing programs or describe adjacent work. They are not extra adoption counts.
Shell’s pump analysis stalled until an operator joined the team
Shell describes an initially unsuccessful search for pump-disruption precursors at Perdido. A field operator helped the data team refine its code and find a chemical signature.
The 70% is a reported precursor association in the analyzed events, not demonstrated live prediction accuracy. Event count, false alarms, held-out validation, intervention results, uptime benefit and full costs are undisclosed.
Limits: The transcript does not establish a live alert or control workflow, prevented disruptions or realized savings. The two-day example is not an average warning time. Initial stalled analysis is not proof of an abandoned or failed field deployment.
The Data Detectives — Solving the 10-year-old problem at Perdido · Unknown
Q3 call: what we checked
Full Q&A caption screen · call 2026-10-29
Complete official Shell Q&A caption export topically screened from 0:12 through the 55:18 call close, including AI/digital, software, automation, pilot, scaling and failure terms; matching passage read in context. Earlier five-page prepared-speech screen retained. Caption-based screen, not audio replay listening or word-for-word accuracy validation.
In response to a refinery-maintenance question, Wael Sawan describes using data and AI to anticipate problems, intervene earlier and coordinate people. He also emphasizes maintenance investment and closer integration of trading with assets.
No isolated AI benefit is quantified. The separately mentioned 99% year-to-date utilization reflects wider refining-team performance.
Limits: No named model, AI deployment site, population, baseline, cost or ROI. The statement that more remains to do concerns broader refining/manufacturing work; it does not establish a failed AI deployment. Approval and control permissions are unspecified.
Shell Q3 2026 results Q&A — official Shell video and captions · Official Q&A captions 0:12–55:18 through call close; refinery answer 21:00–23:44, especially 23:07–23:39.
Read the call record and coverage notes