Your Reservoir Model Came With Baggage. CLARISSA Unpacked It.
CLARISSA reproduced an inherited reservoir model before accepting edits. A Road to ATCE story about dependencies, verification, and engineering confidence.
Print dispatch →Printout № 04 — Dispatch
Unit BW‑26 · 300 baud · ok
Job card — Blauweiss AI
Dispatch archive · verified
Technical essays and research notes from the team building CLARISSA, PetroScript and RIGOR.
New to agentic reservoir simulation? Read how AI builds a reservoir simulation model, then explore the CLARISSA technology stack and OPM Flow workflows. For evaluation criteria, see how RIGOR scores an agent.
CLARISSA reproduced an inherited reservoir model before accepting edits. A Road to ATCE story about dependencies, verification, and engineering confidence.
Print dispatch →Frontier AI models can increasingly run reservoir simulation workflows. CLARISSA and RIGOR focus on the harder problem: verifying that the engineering is correct.
Print dispatch →Why CLARISSA must prove its value through reservoir-specific verification, reproducibility, and a direct comparison with raw Opus + OPM Flow.
Print dispatch →A practical look at AI reservoir simulation, deterministic verification, OPM Flow execution and engineering read-back.
Print dispatch →Why reservoir simulation is a powerful proving ground for verifiable scientific AI.
Print dispatch →Why engineering verifiers and useful approximations matter more than headline benchmark wins.
Print dispatch →On containment, reward functions and the engineering control-room problem in AI systems.
Print dispatch →How RIGOR evaluates reservoir-engineering agents with OPM Flow tasks and verifier-grade scoring.
Print dispatch →On memory, custody, compliance and local deployment when AI enters the asset team.
Print dispatch →· · · End of Printout · · ·