How would you assess an AI-built geosteering tool before using it on a rig?

November 19, 2026 5 answers
Asked by _h26+brisketuno.5

A contractor has built a geosteering application with AI coding assistants and wants to charge $500 per month per rig. It imports well logs and surveys for 2D cross-sections, with a real-time data feed still in development.

We work as separate contractors, and the tool is being tested on jobs. What should we check before relying on it, and how should we think about liability and the proposed fee?

Answers

Ruvmelo.5f

Keep it out of decisions that control the well until someone independent has validated it. A parallel trial against an established workflow is useful, but matching another product alone does not prove either result is right.

Start with cases where the expected geometry and calculations can be checked independently. Test units, coordinate conventions, survey interpolation, depth alignment, missing or malformed LAS data, and edge cases. Compare results on historical wells too, and investigate differences instead of assuming the familiar tool is always correct.

Have a qualified geosteering specialist review the geological assumptions and someone with relevant software expertise review the calculations and tests. Record the version, inputs, settings, outputs, and known limitations. AI reviewing AI-generated code is not independent validation.

Ravtuno-3k:

The real-time feed needs its own tests too. Delayed, duplicated, incomplete, or out-of-order data could make a plausible-looking display misleading.

Reply to this answer
IbisSwitch

Separate testing from operational authority. Agree with the operator on which tool and workflow are approved for decisions, who can approve a change, and what happens if the new application fails or produces a different interpretation.

Keep a usable fallback and make sure the people making drilling decisions know which data and interpretation they are seeing. A passive trial can still create confusion if someone treats its output as authoritative, so label it clearly and document discrepancies.

Reply to this answer
Zavqueno-7x

The monthly price is the easy part. What are you buying for it? Get the licence terms, support hours, bug-fix process, data handling, update policy, and a way to export your work in writing. Also ask what happens if the developer stops maintaining it.

A developer getting angry about bug reports is a practical concern, not just a personality issue. You need a way to report defects, track their severity, and know when a fix has been tested before anyone depends on an update.

Reply to this answer
5d~fablecompass_613

Have a lawyer familiar with your contracts and jurisdiction review this before signing or relying on the software. Separate contractor companies do not automatically settle who owns the code or who bears a loss. The development circumstances, agreements, and any operator requirements matter.

Ask about liability limits, indemnity, and insurance coverage for the actual work and software use. An indemnity is only as useful as the party's ability to honour it. Also have counsel examine the commercial vendor's terms and what was actually copied or used. A possible breach is not something the discussion can resolve from a description alone.

Reply to this answer
_k39~torvani.6

AI assistance is not by itself proof that the software is unsafe, and an established product is not automatically bug-free. The question is whether this particular tool has evidence behind its calculations, dependable maintenance, and controls appropriate to the decisions it supports. Field experience can help identify useful features, but it does not replace those checks.

Reply to this answer

Add your answer

Answers are reviewed before publication.

Related Discussions