What I Learned Teaching AI Across Four Sectors

When I started doing AI training seriously, I thought the sector wouldn't matter much. The tools are the same. The models are the same. Surely the session content would be 80% identical with a few industry-specific examples swapped in.

I was wrong.

After training teams across banking, insurance, real estate, and agribusiness in Morocco, I've come to believe that sector shapes everything, not the technology, but the fear, the urgency, the resistance, and the specific breakthrough moments that make the training actually land.

Here's what each one taught me.

Banking: the compliance wall

Banking executives are not slow to adopt AI. They're careful. There's a difference.

In every banking session I've run, the same question comes up within the first thirty minutes: "Where does our data go?" And they're right to ask. These are institutions that operate under strict regulatory frameworks, that handle sensitive client information, that cannot afford a data breach or a compliance failure.

The answer they need isn't "don't worry, it's secure." It's a technical explanation of local deployment: models running on their infrastructure, no data leaving their perimeter, full audit trails. When I explain that a language model can run entirely on-premise, that the AI processing their client queries never touches an external server, the wall comes down.

What unlocks for banking teams: customer support automation is the obvious one, but the deeper opportunity is internal. Document generation, regulatory compliance checks, meeting summaries, internal knowledge bases that actually answer questions instead of returning a 200-page PDF.

The best moment I had in a banking session was when a department head realized she could build something that answered the twenty most common questions her team gets from junior employees every week, not by training a chatbot, but by connecting a language model to their existing documentation. Two hours to understand the concept. One week to prototype it. That's the banking pace, and it's actually the right pace.

Insurance: the distribution problem

Insurance in Morocco has a specific challenge that AI is uniquely positioned to solve: the gap between what agents know and what clients need to understand.

Insurance products are complex. The language is technical. The conditions are long. And the agents explaining these products to clients are often under time pressure, juggling dozens of clients, working from memory. The result: inconsistent explanations, missed details, clients who sign contracts they don't fully understand.

What I proposed in every insurance session was a shift from AI as an internal productivity tool to AI as a client-facing communication layer. A system that takes a complex insurance product and explains it in plain French (or Darija) based on the specific profile of the client. Not a generic FAQ. A contextual, personalized explanation.

The reaction in these sessions was always the same: immediate recognition. Every insurance professional in the room knew exactly the problem I was describing, because they'd lived it. That recognition is what makes the training land. I'm not introducing a foreign concept; I'm naming something they already know is broken.

Real estate: the content bottleneck

This is where AI has had the most visible, immediate impact in Morocco.

Real estate developers here produce enormous amounts of content: listings, floor plans, lifestyle photography, client communications, promotional videos. All of it was either expensive to produce or slow to produce or both. A photoshoot for a residential project could take half a day, require a photographer, a stylist, models, post-processing, and still come out looking generic.

What changed is simple: we can now generate lifestyle photography that doesn't look generated. An apartment that exists only as a 3D render can have a living room photo that looks lived-in, warm, real. We can produce twelve versions of the same visual for different client profiles without twelve shoots. We can generate a complete property listing from a briefing in thirty seconds.

When I show this to a real estate developer (not explain it, show it with their actual project), the conversation stops being about AI and starts being about their pipeline. How many listings do they have to produce? How many campaigns are they running? What would it mean to cut production time by 80%?

That's the real estate unlock. It's not about replacing creativity; it's about decoupling content production from calendar and budget.

Agribusiness: the document problem nobody talks about

This one surprised me.

Agribusiness companies in Morocco are not, on the surface, obvious AI candidates. They're dealing with supply chains, seasonal production, physical goods. Not knowledge work in the traditional sense.

But here's what I found: they're drowning in documents. Certificates of origin, phytosanitary documents, compliance reports, supplier communications, product specifications. Every export shipment generates paperwork. Every regulatory change requires updating a stack of documents. Every new client needs a tailored presentation of capabilities.

All of this is currently done manually, from scratch, every time. And all of it is exactly what a well-configured AI system handles well.

What landed in agribusiness sessions was the document automation angle: not chatbots, not image generation, but the mundane, non-glamorous work of taking structured information and turning it into formatted documents. Fast. Consistently. Without errors from copy-paste fatigue.

The secondary unlock was sales enablement: training commercial teams to use AI to research prospects, personalize proposals, and follow up intelligently instead of sending the same generic email to everyone.

What's universal

One thing doesn't change across any sector: the people who get the most out of AI training are the ones who come with a specific problem, not a general curiosity.

The executive who walks in saying "I want to understand AI" leaves with an interesting afternoon. The executive who walks in saying "I have a specific process that takes my team three days every month and I think it shouldn't" leaves with a prototype plan.

Sector shapes the problem. The mindset shapes the outcome.

That's what four industries taught me.