Why I’m Experimenting with local AI

I live in Europe.

Most of the AI tools I use every day come from the United States.

They’re powerful. Fast. Incredibly useful.

But something is becoming clearer.

If you work with sensitive information—client strategy, internal discussions, early-stage ideas, confidential documents—at some point, you start asking a different question.

Not “How good is the model?”

But “Where is my data actually going?”

For many use cases, cloud AI is perfectly fine.

But for others, especially in advisory or leadership work, sending sensitive information into external systems becomes harder to justify.

That’s why I’ve started experimenting with offline and local AI setups.

I mean, not because they’re better. Often they’re NOT. But because they offer something increasingly valuable: control.

Here are three directions worth exploring if confidentiality matters in your work.

1. Local models running on your machine

Tools like Ollama or LM Studio allow you to run open-source models directly on your computer. No data leaves your device. You trade some power for privacy and control.

2. Self-hosted models

More technical teams can run models on their own infrastructure. This allows companies to keep sensitive workflows inside their own environment.

3. Hybrid workflows

The most realistic approach for many people:

Use powerful cloud models for general work, and local models when dealing with confidential material.

AI is evolving quickly, but the next phase of the conversation isn’t just about capacity and power. It’s sovereignty, control, and trust.

Especially if you work with information that cannot leave the place where you share it.