Zero to Hero AI implementation strategy — An EU-based hardware company

For the last 5+ years, everyone’s been talking about FOMO—fear of missing out. It kicked off with crypto and has rolled straight into AI. Now the chatter is all about how easy AI is, how everyone’s using it, and how if you don’t have 50 AI agents working for you, you might as well close up shop because your AI-infused super-competitor will eat your lunch.

Normally, it’d be easy to call bullshit on broad claims like that. But this time, the tech is actually that good as well as accessible so there’s a grain of truth. Look closer, though, and it’s the same pattern you see with every new technology: everyone talks, many tinker, few get real value. Making AI work for YOUR company still takes investment, thought, and a lot of experimentation.

That’s exactly where an EU-based hardware manufacturer came to Subsense. They’d sold equipment successfully for years, but competitors with stronger software and AI features were gaining fast. The goal was simple: shift from a hardware company to an AI company. Sounds like you can just drag-and-drop your docs into ChatGPT, right? Not even close—unless you’ve got the budget of OpenAI or Microsoft.

To use AI properly, the basics still apply:

Data is king. If you’re not generating and collecting data at every node—and making it accessible in a single source of truth—any sustainable AI future for your company is impossible.
AI belongs next to the product, not buried under the CTO. Yes, AI is technical. No, it shouldn’t live in a back room. It has to be embedded in the product context and close to the business.
Build in-house talent. You need people who understand your data, your product, and your customers.
Keep it simple, stupid. Complexity is the enemy of shipping.
Measure and guardrail constantly. If you’re not tracking performance and risk, you’re guessing.

That list is in order of importance. Fancy models aren’t the differentiator anymore. How your company’s data plugs into modern LLMs and VLMs is.

Okay, the case study. We went from zero AI to a working ecosystem—with real bottom-line impact—in three months. We:

Turned on the data taps. We started capturing data at every generation point and set up Databricks as the single source of truth for the whole company.
Hired in-house. We built a data science team reporting directly to the Head of Product.
Picked our shots. Using the new pipeline, we identified 5–6 high-leverage areas for AI agents. We shipped agents that automated in-house manual work, scraped the web for new sales opportunities, and integrated directly with the product.
Made automation someone’s day job. One senior developer owned continuous automation. It didn’t have to be AI-related—just relentless.
Built monitoring and MLOps from day one. Whether an agent adds value should never be an executive’s gut feel. It’s a metric. We instrumented everything.

The journey doesn’t end there—it starts there. But that first wave gave the company a real jump-start and visible value, which then began to spread on its own. Every company is different, and what worked here might not map 1:1 to yours.

If you want to figure out the best AI strategy for your company, get in touch.