AI consultant and architect for industry

Castellano · Català

I'm Cinto Casals. I help manufacturing companies in Barcelona, Catalonia and the rest of Spain turn AI into something that works on the shop floor: a corporate memory that lives on your server, written rules on who decides what, and agents connected to real processes. I don't sell licenses. I design the architecture and take it into production.

In one sentence: if you are looking for an AI architect or an AI advisor who knows the shop floor, it starts with a three-week diagnosis, the Paso Cero (€3,500 + VAT), and if you decide to implement, we take the AI into production in three to four months, with a written completion criterion for every phase.

Who it is for

For general management, technical management and operations management in companies with a plant, an ERP and real processes. For people who already know that AI requires data engineering and are willing to blow up legacy processes instead of automating them as they are.

It is not for people looking for prompt tricks, or for those who believe AI installs with one click. If your AI strategy is handing out ChatGPT licenses, you are not innovating: you are automating chaos.

Why someone who has worked on the shop floor

I have been building digital infrastructure since 1996. I founded Interpisos, the first real estate portal in Spain, and ran Rentadata Networks as CEO and CTO from 1998 to 2012. Later I spent more than five years in production, in industrial printing companies, with the ERP, Excel and machine systems that don't talk to each other, and with day-to-day quality, materials and incidents.

That combination is what matters in a factory: knowing how to build systems and knowing what happens when a machine stops at three in the afternoon. AI that doesn't fit into that flow stays a demo.

What your company will have at the end

  • Memory. A single place where what your company knows lives. It answers and says which minutes, email or report each answer comes from.
  • Criteria. Written rules: who can ask what, what leaves your server and what doesn't, and who is accountable for what a machine decides.
  • Execution. Agents connected to real processes, each with a named owner and a measured saving.

How I work: five phases, each with its own completion criterion

  1. Phase 0Paso Cero. Know what is there before touching anything.3 weeks
  2. Phase 1Cortex. The memory, on your server.3 weeks
  3. Phase 2Direction. An AI that knows the business.3 weeks
  4. Phase 3Workforce. Agents on real processes.4 weeks
  5. Phase 4Autonomy. So you don't need us.3 weeks

I do it through SerIA Nativa, the consultancy I set up to bring the AI-Native Company to industry. The method and the rates are on serianativa.com (in Spanish).

Where it starts: the Paso Cero diagnosis

Three weeks to find out what AI can do in your company. And what it can't.

  • Shadow AI inventory: which tools are used, who uses them and with what data.
  • A map of critical knowledge and of how many people hold it.
  • The three candidate processes, scored and with their estimated return.
  • Target architecture: what lives on your server and what leaves it.
  • A four-month plan with milestones, owners and figures.
Price
€3,500 + VAT
Duration
3 weeks
If you sign up within 60 days
Fully deducted

Completion criterion: you can explain the plan to your board in five minutes. If you can't, we're not done.

I want the Paso Cero diagnosis (page in Spanish). Three weeks, €3,500 + VAT, and a plan your board understands in five minutes. →

Frequently asked questions

What does an AI architect do in a manufacturing company?

Decides which AI is worth building, where each piece of data lives and how it connects to the ERP, the machines and the people. Then takes it into production with an owner and a measure for every use. Doesn't sell licenses.

How much does it cost to start?

The Paso Cero diagnosis costs €3,500 + VAT and takes three weeks. If you sign up for the implementation within 60 days of receiving the report, it is fully deducted. The implementation is quoted once the diagnosis is done.

Do I have to replace my ERP?

No. The point is to integrate AI without breaking operations: to connect the information that today is scattered between the machine, the ERP, Excel, paper and people's heads.

Does my data leave the company?

The corporate memory lives on your server. What leaves and what doesn't is written down in the diagnosis, before anything is built.

RAG or fine-tuning for ERP data?

Almost always RAG: a memory that queries your data and cites where every answer comes from. Fine-tuning is expensive and ages every time your data changes. I explain it in detail in the manual, steps 2 to 5.

How do I know if I need a Paso Cero?

Answer the five questions in the one-minute test. If any answer is "no", the Paso Cero will tell you where the real return is. With three "no" answers or more, you are automating chaos, even if it doesn't show yet.

The book. “La Empresa IA-Nativa” (in Spanish): the architecture in 11 steps, paperback and Kindle, with a free summary on the site. If you'd rather do it yourself, this is the blueprint. →
Want to see how it works before we talk? My system, live (in Catalan): the projects and agents I work with every day. Would you rather have a talk for your management committee? Talks. A specific question? Write to me.

Keep reading

Cinto Casals, AI architect for industry (Castelldefels, Barcelona). Founder of SerIA Nativa and author of the manual “The AI-Native Company”.