Home / FAQ

Frequently asked questions

The questions I hear most often about AI in SMBs, Quebec's Law 25 and mining data. If yours isn't here, just ask me.

AI in your business

Where should an SMB start with AI?

With a conversation, not a tool. I come to your office, at no cost, to understand how work gets done and where it gets stuck. We often find that your employees are already using AI. We start from there to pick a few useful use cases, rather than launching a big project.

My employees already use ChatGPT or Copilot. Is that a problem?

Not in itself. It's actually a good sign: they see value in it. The risk comes from personal accounts and client data pasted into a consumer tool without the company knowing. The answer isn't to ban it. It's to set clear rules and provide the right tools.

Can I put client data into ChatGPT, Copilot, Claude or Gemini? What does Quebec's Law 25 change?

It depends on the tool, the version and how it's configured. A properly set up business version has nothing in common with a free personal account. As soon as a tool processes personal information, Law 25 applies. A new system that processes it often requires a privacy impact assessment (PIA), and most AI tools process data outside Quebec, which needs to be assessed before you entrust them with personal information.

I'm not a lawyer. When a question becomes legal, I tell you and we bring in your advisor. See AI governance and Law 25.

What's the difference between AI and automation?

Automation does exactly the same thing every time, following fixed rules, for example moving invoices from an email into your accounting software. Generative AI interprets, writes and summarizes, with some variation that a person needs to check. Often, the right answer to a pain point is a simple automation. I'll tell you honestly which one you need.

Which tool should we choose if we already use Microsoft 365?

Microsoft 365 Copilot has the advantage of working in your documents, email and Teams, while respecting the permissions you already have. It isn't always the best choice for every task: ChatGPT, Claude or Gemini sometimes do better for certain uses. I look at your real needs before recommending anything.

Do you sell licenses or software?

No. I don't sell licenses and I have no agreement with any vendor. License costs are confirmed with your IT provider or reseller. My only interest is that the tool you choose actually serves your team.

Is training in person or remote?

One-on-one training happens in person, at your office, because that's where we work on your real tasks and documents. For the rest of a mandate, part of the work can be done remotely.

One-on-one or group training: which should we choose?

Group training builds a shared language and moves the whole team forward at once. One-on-one goes further into one person's tasks. The two work well together: a group session to get started, then one-on-one for the people leading the change. We choose based on your situation.

How much does it cost?

Each mandate is priced to fit, after a free first meeting. I then give you a written proposal with the scope, deliverables, timeline and price, before anything starts.

Where do you work?

Anywhere in Quebec. I travel for meetings and training, and part of the work can be done remotely depending on the mandate. I work in French and English.

How long does a mandate last?

It depends on the goal. A training session or a talk is a one-time engagement. Governance or an adoption strategy runs over several weeks, sometimes several months, with spaced-out follow-ups. The timeline is written into the proposal.

What happens when the mandate ends?

The goal of every mandate is your autonomy, not your dependence on me. I identify and equip internal champions, and I leave you documents your team can keep alive. If you want follow-ups, we plan them, but they're not required.

Mining data

How does the discovery phase of a mining mandate work?

I start by understanding your vision, your challenges and your data sources: who produces what, who uses it, where things get stuck. I then deliver a picture of the current situation and a prioritized game plan that fits your team. Further mandates follow from it, if you want them. See Mining data.

What does "AI-ready data" mean?

An AI assistant is only as good as the context it's given. AI-ready data is structured and documented: definitions, sources, units, history. That's what lets generative AI query it and produce reliable answers. It's often the condition for an AI project at a mine to deliver something concrete.

How do you stay independent from platform vendors?

I don't sell any platform and I have no agreement with any vendor. I help you define your needs, sit in on vendor demos with you and ask the questions vendors would rather avoid. You come away with a reasoned recommendation whose only interest is yours.

Which mining data platforms do you know?

The mining ecosystem is broad, and a good part of the work is getting these systems to talk to each other. Here are the ones we run into most often:

  • Drillhole and geology data: acQuire, Seequent (Leapfrog, MX Deposit), Datamine
  • Mine planning: Deswik, Hexagon MinePlan
  • Fleet management and dispatch: Modular Mining DISPATCH, Cat MineStar, Sandvik OptiMine, Wenco, HxGN MineFleet
  • Process data (historians): AVEVA PI System
  • Laboratory (LIMS): CCLAS, LabWare
  • Environment: EarthSoft EQuIS
  • Maintenance: SAP PM, IBM Maximo
  • Data warehouse and BI: Microsoft Fabric and Power BI, Snowflake, Databricks

The list isn't exhaustive, and I don't recommend any of them in advance. The right choice depends on your needs.

It starts with a conversation.

I come to you, at no cost, to understand your situation before proposing anything.