Two in-person half days at Italy's first AI-native venture studio. You learn to adopt AI in your company and keep control of your data, your models and your suppliers.
Run by the team that builds KVA's own AI products.
The workshop is two half-day episodes at KVA's studio in Turin. It is about Sovereign AI. We show how a company adopts AI and keeps control of its data, its models and its suppliers. Each episode opens with a short part on the choices involved. For the rest of the morning you work on your laptop with our team beside you.
KVA is Italy's first AI-native venture studio. For two years we have shipped AI products for our own portfolio. We also help industrial groups adopt AI. In the room we bring the tools we use every week and the mistakes we have already made.
Two episodes before the end of the year: Friday 20 November and Friday 18 December, from 9:30 to 13:30. Each episode stands on its own. Eighteen seats per episode, at € 199 + VAT. We work in Italian or in English, depending on who is in the room.
Each one applies to a project already on your desk.
After the first episode you can place any AI offer on four axes: data, model, infrastructure and skills. You can tell when a supplier asks for more access than it needs. The same map works for the products that come out next year.
In the first episode you build an assistant on documents from your own work. Then you swap the model underneath and compare the answers. You see what it costs to change supplier when the architecture is done well.
You leave the two episodes with the questions to ask before you sign. They cover costs, data, security, contracts and the AI Act. You use them to judge proposals from suppliers, consultants and your own team.
Two half days, each with its own agenda. Each episode stands on its own, so you can join one or both. Each block is led by one person from our team. Coffee is included.
Bringing AI into a company and building a product.
We start from the AI market and how it moves. We look at where AI brings measurable benefits to a company and what it really costs. We choose where the data lives, which model to use and where to run it. We close with the risks and with what the AI Act asks.
You bring a laptop and a few non-confidential documents from your work. We set up the environment together and we pair with you. You build a small assistant that answers and cites its sources. Then you swap the model and see how the answers change. Claude Code writes the code from what you describe. You leave with a working prototype and a repo that stays yours.
Protecting AI in the company and using it to optimise.
AI opens new doors to attackers. We look at how text hidden in a document can hijack an agent. We look at how data leaves a company through an assistant. We look at how attackers use AI for phishing and fraud. For each threat we show a concrete defence and the questions to ask suppliers.
A model is useful when its answers rest on verifiable data and results. We look at how to get that grounding. We look at how AI helps solve optimisation problems. We look at how to measure the ratio between tokens spent and results obtained.
You work on a ready-made assistant. First you try to attack it with the techniques from the security block. Then you protect it. Finally you measure what each answer costs and how reliable it is, and you try to improve it. We close with questions at 13:30.
Federico, Alberto, Carlo and Davide each lead the block they work on every day.
Managing Director, KVA · Faculty, ITS Torino
Opens the first episode. Explains how to bring AI into a company and what it costs. Covers benefits, risks and how the AI market moves.
KVA · AI products
Leads the hands-on part of the first episode and pairs with anyone who needs it. Built the agent infrastructure KVA uses every day. Explains how to build a product with AI.
Co-founder, KVA · Cybersecurity
Comes from offensive cybersecurity. Explains the threats AI brings into a company and the ones it makes easier. Shows how to protect data, models and agents.
Theoretical physicist · KVA Scientific Committee
Uses AI to do science. Explains how to ground a model in verifiable data and results. Shows how AI is used on optimisation problems and how to measure what each token spent returns.
Each episode is a half day and stands on its own. You can join one or both.
Federico covers the benefits, costs and risks of AI in a company and how the market moves. Then Carlo leads the hands-on part, where you build an assistant on your own documents. It is designed for CIOs, CTOs and heads of innovation and product.
Alberto covers the threats AI brings into a company and how to defend against them. Davide shows how AI is used to do science and to optimise. Then you put an assistant to the test. It is designed for heads of security, risk, data and operations.
Two recent examples, in the words of the people who shipped them.
I joined KVA from a more traditional consulting background, expecting to keep doing what I'd done before, running strategy projects, writing reports, presenting findings. Instead I ended up building things I would have outsourced in any other job: prototypes for clients, internal tools the team actually uses, a dashboard I shipped in an afternoon to track our portfolio. What surprised me most was how quickly the gap closed between having an idea and having a working version of it. I still write strategy when it's needed, but now I can also build the thing the strategy is about.
I joined KVA as a writer, and before long I found myself doing things I never thought I'd be able to do: building websites, working with CMSs, shaping brands, handling deploys. What surprised me most was how naturally I picked up tools like Claude Code, GitHub, and Vercel by working directly on real projects. To me, vibe coding is about speeding up your work by learning as you go, where the real limits are less and less about technical specialisation and more and more about vision, taste, and your ability to be surprised by what you can build.
At checkout you can book several seats for your colleagues in one purchase, with one invoice. We can also bring the workshop to your company, as a half day or a full day.
If yours isn't here, write to us, we answer in a day, usually the same afternoon.
Eighteen seats per episode, with the tools we actually use.
18 seats per episode · € 199 + VAT, students and researchers € 99 + VAT · Full refund up to 7 days before
