Skip the six-month pilot. We pick the one place where AI saves your team the most time, and in under ten days your people are using it. Then we decide what comes next.
From kick-off to live · 10 days
Day 1
Kick-off.
The scope is already agreed. We connect to your tools and your real data.
Day 2–9
Build.
Built and tested on your real work, inside the tools your team already uses.
Day 10
Live.
Your people are using it. Then we decide what comes next.
The 10 days start once the project is defined. First we scope it together on a free call: what to build, for whom, and what done looks like.
“Very nice work, was quick and thorough… understood requirements perfectly.”
5.0/5client rating
Ten days to a first version. Then it keeps paying for itself.
Two ways to work together
For companies
Documents to check, calls to answer, tickets to sort. I build the AI that does it inside the tools you already use, and keep it running.
For agencies & software houses
When your team is full, a project is late or a client asks for AI, I join as one more senior, under your name if you prefer.
What I build
Work your team repeats every day, like sorting, copying and checking, done by an agent inside your tools.
e.g. 1M+ requests a year sorted automatically
Answers from your manuals, procedures and catalogue, with sources, checked before launch.
e.g. 2M+ products searchable by just asking, every item checked against the request
Replies and bookings at any hour, with a person stepping in when needed.
e.g. 0 missed bookings in month one
An audit and rewrite of the AI you already run: same results, for less.
e.g. −80% AI costs with the same features
Work
Banking
1M+tickets / yr
LLM agent sorting into 400+ categories, 81% accuracy vs 75% human.
Banking
−80%handling time
OCR, signature verification and validation rules for a major Italian bank.
Banking
2,500+docs / day
Microservices pipeline with audit trail, 84% less manual work.
Healthcare
0missed calls
40+ doctor clinic, 96.7% success score in the first month.
Retail
90%+fully correct
Hybrid vector search over a catalogue of 2M+ products.
Fintech
−80%AI costs
Consumer finance and investing platform, same features, far lower spend.
Shipped for banks, clinics and fintech, where a mistake costs real money.
Testimonials
★★★★★
“Great to work with. Went above and beyond to achieve our goals.”
★★★★★
“Very nice work, was quick and thorough… understood requirements perfectly.”
LinkedIn Reference
“If you need someone who consistently delivers top-quality results, Giacomo is the ideal choice. I collaborated with him on a highly complex project. He tackles difficult challenges with remarkable dedication. His deep expertise in his field is a plus to the team. He communicates effectively and drives team progress and I would eagerly work with him again.”
FAQ
First version from €4,900, fixed price agreed after the free scoping call. Then maintenance from €490/month, or new phases quoted separately.
Yes, an NDA is standard. Your data stays in your environment or in an EU cloud you control, and models are chosen with GDPR in mind. Most of my enterprise work is confidential for exactly this reason.
My go-to stack is FastAPI, Next.js and TypeScript, PostgreSQL, S3 and Cloudflare, plus LiveKit for voice agents. It covers almost every project I take on. I've worked with other setups too, so with agencies I work in your stack and your repository.
You have something running on real cases and real numbers. From there we decide together: harden it for full production, extend it to the next workflow, or stop. Support and maintenance are available after go-live.
Yes. I'm based in Italy and work remotely with teams in Italy and abroad, in Italian or English.
About
Master's in AI & Data Engineering from the University of Pisa. Years building generative AI for banks and insurers, now leading AI on products that real customers use every day. I take a system from use case to production, and stay for what comes after.
More
