Giovanni Liguori, AI Automation Architect.
I train teams to use AI in their work and build agentic systems that work on their documents. For companies and service firms that run on case files, reports, procedures and email.
Where the method comes from
I build AI systems that turn manual processes into reliable, observable and measurable operations.
I work with companies, service firms and professionals, starting with the problem, not the tool: I identify the bottleneck, map the process and define what can be automated and what has to happen when something goes wrong. Only then do I choose models, tools and infrastructure.
Actually, I didn’t get here by starting with AI.
My background has always been technical: Linux, programming, C++ and Python. I was interested in understanding how systems worked, and in building, modifying and automating them.
Professionally, though, I took a different path and spent years working in video editing. It was an important craft for me, but over time I realized I had drifted away from the kind of problems I found it natural to tackle.
I was interested in understanding how systems worked, and in building, modifying and automating them.
Even while I was working on timelines, audio, rendering and deliverables, I kept asking myself: why am I doing this by hand? Why don’t these steps talk to each other? What can I eliminate, automate or make more reliable?
I started by automating individual tasks. Then entire processes.
The arrival of AI models didn’t create that interest: it gave me a new component for building much more capable systems.
Today my own business runs on a hybrid ecosystem of agents, traditional software and automations that coordinates content, analysis, monitoring and operations.
My main stack is Claude and Python, but the model is only one part of the system. I separate probabilistic work from deterministic work and design runtime, state, tools, fallbacks, gates and alerts so that AI can operate autonomously without turning into a black box.
It’s also how I work with clients: I don’t start with the question “where can we put AI?”
I start with a more useful question:
how much manual work can we eliminate without losing control?
I don’t sell miracle prompts or automations built around the hype of the moment. I build systems that have to work even when the model gets it wrong, a service doesn’t respond or something goes off the expected path.
Tell me about your team. Let’s decide together where to start.
To train your team, start with the program and the questionnaire. I’ll reply with a proposal tailored to your people, goals and tools.
For a technical project, the scoping call takes 30 minutes and costs €97, deducted from the quote. If the process is already chosen, the 90-minute review costs €297 and defines its boundaries before the build.
Training, reviews and agentic systems can all be delivered in English and remotely.