6 February 2026 · Pedro Aldea
Why we talk about operations, not AI
AI doesn't fix broken processes. Operations come first. Then, if it earns its place, artificial intelligence follows.
When I say we’re an operations team and not an AI consultancy, people get confused. “But you do AI…” Yes. We do AI. But we don’t start there.
The problem with starting with AI
Most companies that “want AI” have an operations problem, not a technology problem. They have invoices processed by hand. Data living in someone’s head. Catalogs nobody has cleaned in a decade.
Putting AI on top of a broken process is putting a Ferrari engine in a car with no wheels.
Our hierarchy
- Eliminate: Does this step need to exist?
- Simplify: Can anyone on the team do it?
- Automate: Can a deterministic rule handle it?
- AI: Does it require judgment at scale?
AI is the last resort, not the first. And it works spectacularly well when applied on clean operations.
A real example
At Grupo Cartes, before building the invoice OCR system, we mapped the entire document processing flow. We discovered that 30% of the steps were unnecessary. We eliminated them. Then automated what remained. And only where intelligence was needed (interpreting unpredictable supplier formats, detecting statistical anomalies) did AI enter.
Result: 95% time reduction. But it wasn’t AI alone. It was clean operations + AI where it earned its place.
The takeaway
If someone sells you “AI” without first talking about your operations, be skeptical. Technology is a means. The goal is for your business to work better, cost less, and scale without pain.
That’s what we do. Obsessive operations. AI is the consequence.