The Board Guide to AI in Mining

FREE REPORT. Does your board know how much AI is already in use inside the business? Do you know the right questions to ask?

Microsoft and LinkedIn's survey of 31,000 knowledge workers found that 78% of people using AI at work bring their own tools. In mining, the drilling contractors, laboratories and consultants who produce much of what reaches the board are making their own AI decisions too, often on the company's data.

Pulse Intelligence created a comprehensive report, including 22 questions, across five key categories. It's written in simple, practical language, not tech hyperbole.

Across 160 client conversations since April last year, the pattern has been consistent: Boards are told what management is doing with AI. They are rarely (yet) asked to direct it, and seldom given the terms to do so.

So we wrote The Board Guide to AI in Mining. It's a comprehensive guide, written for chairs and non-executive directors, C-suite executives and leaders in mining finance.

Two things in it I would push any board on.

Capital has already picked a direction.

KoBold Metals at a US$2.96bn valuation and Mariana Minerals at US$1.5bn are priced on the AI thesis, much of it still ahead of proof. If that premium holds, the companies with the data foundation raise on better terms. If it fades, it inverts into a discount for whoever arrives without one. Either way, the laggard pays.

Secondly, it's clear that the binding constraint is the material the model is given - e.g., structured vs. raw data. Gartner expects 60% of AI projects without AI-ready data to be abandoned through 2026.

On our own benchmarking, the same decision-grade question takes about 5,000 tokens on structured, validated data and 1,000,000 or more when a model re-reads raw scans. Most pilots make this worse because each one extracts the same documents again and leaves nothing behind for the next.

The resource statement four tools touched

The guide works one case through in full. A mid-tier gold producer is updating its resource statement. The drilling contractor's core-logging software classifies rock types from core photographs. The laboratory screens its quality-control results with an automated system. The consulting geologists draft sections of the technical report in a general-purpose AI tool, pasting in extracts from the company's confidential drill database. The investor-relations agency drafts the market announcement with another.

None of it appears on the risk register or in the company's contracts, and no one can say which tools touched the numbers in the final release. The Competent Person still signs, and the company still owns the disclosure. When a contractor's AI makes an error, it becomes the company's disclosure.

22 questions for the board

The guide closes with 22 questions for the board, starting with the basics and then grouped around capital allocation, risk, disclosure and control, and people, and a plain-English lexicon so directors can test what they are told. Three of them:

• Do we know where AI is already used in our business? Is there an inventory, and who keeps it current?
• Do our contracts cover AI? Must contractors and advisers tell us when they use it on our data, and how they protect that data?
• Do the people who sign know what AI has touched? Do our Competent and Qualified Persons know which tools shaped the numbers they sign?

Get the guide, free

27 pages: where the industry stands, where it is going over the next decade, risk and control, the AI premium in valuations, the data foundation, talent, where to start, all 22 questions and the board lexicon.

Enter your email and we'll send the PDF straight over.

Less searching. More strategising.™

---

General information for boards and directors, not investment, legal, technical or financial advice. Third-party figures remain their authors' work; survey figures are respondents' own reports; token comparisons come from Pulse benchmarking in June 2026 and are illustrative rather than audited constants.