Context
A 15 million-tonne-per-annum refinery wanted one trusted, live view of operations.
Challenge
Operational data was spread across dozens of IT and OT systems.
What we deployed
An enterprise digital twin with Refinery-GPT and operational agents.
Typical architecture
- STEP 01
Historians, LIMS, ERP and maintenance systems connected
- STEP 02
Deterministic twin computes every number
- STEP 03
Refinery-GPT answers questions over the twin
- STEP 04
Agents draft reconciliations and handovers for review
Illustrative of how this solution is typically structured and governed; deployment specifics are available under NDA.
Results
- 15 MMTPA refinery modelledPartner-reported
- 5 people on the build teamPartner-reported
- 6 mo to an enterprise digital twinPartner-reported
- Refinery-GPT and operational agents for inventory reconciliation and shift handoverPartner-reported
NoteResults from deployments by GoGenesis and its platform partners. Outcomes vary by portfolio and operating environment; your results are set in the pilot.
Governance approach
- Numbers computed, never inferred by the language model
- Operators approve agent outputs
Illustrative of how this solution is typically structured and governed; deployment specifics are available under NDA.