Five of us, one table

A young Italian company, made of five people who hold together security, product, AI research and community. We build MiniMako to keep AI agents on course, without slowing them down.

Join the waitlist

The team

The people building MiniMako

Five people, five stories held in balance. Security, product, research, business, community.

Dario Vignali

Co-founder · Brand & Community

Built Marketers, Italy's largest community of digital professionals. Made the Forbes Under 30 list in 2020. At MiniMako he leads the brand's voice.

Pietro Virgillito

Co-founder · CTO

Spent twenty years on the operational security of Italian banking and public-administration infrastructure, and works on agentic systems from before they went mainstream.

Francesco Italia

Co-founder · Product & Agentic Engineering

Has scaled applications to millions of users, with distributed-systems experience few Italian startups have really touched. At MiniMako he builds the product and the agentic systems.

Carmen Palmisano

Co-founder · Go-to-Market

Founded her first company at twenty-six. Spent years in corporate IT consulting. At MiniMako she holds together vision, product and market.

Luca Cresi Ferrari

Co-founder · Go-to-Market

Founded and scaled a group of companies to millions in revenue without outside capital. At MiniMako he leads go-to-market.

How the idea was born

We found each other for the same reason you're probably here

Each of us (freelance, product teams, companies) started working with AI agents and noticed the same thing. They give you enormous leverage when they work, and they can do real damage when they go off course. A file deleted over a misread instruction, a token spent where it shouldn't have been, a skill installed that nobody remembered.

The community around this technology was growing fast, and the tools to control it were falling behind. Five of us sat down at the table, from different backgrounds: some from twenty years of operational cybersecurity on banks and public administration, some from years of product scaled to millions of users, some from studying artificial intelligence in the right labs, some from community and digital storytelling in Italy.

That's where MiniMako came from.

How we work

A table where people reason, not a pulpit to preach from

We ask questions, bring real cases, and put together people who think about AI without hysteria and without selling fear. If you read the newsletter or drop by one of our live sessions, you'll find a table before you find a product. Then there's the product, which we build with the same care we put into writing a piece.

Read the newsletter

Why we're here

Controlling an AI agent is an act of responsibility, not paranoia

Each of us has seen, in our own field, what happens when new tools arrive without control. In security (data breaches), in product (features that break systems), in community (people learning the wrong way because nobody explained it properly). AI today is at that same moment: powerful, fast, and without guardians. We want to build those guardians ourselves, in a way that doesn't feel like a cage to the people using it.

The rest is in the product and the newsletter. We'll leave the link below.

Events

Where to find us over the next few months

If you want to meet us in person, here is what we have on the calendar.

Oct 3

RomHack

Villaggio Flaminio, Rome

How to poison a skill, and why nobody reviews it twice

Saturday, October 3, 10:20 AM Stage 2 · Ghost in the Shellcode

What it's about

Nobody runs a bare AI agent. Everyone extends it: skills, MCP servers, plugins, hooks, subagents. Every one of those is, underneath, a folder of text that gets loaded into the model's context and read as instruction.

Which makes them a supply chain. One that gets reviewed exactly once, at install, and executed every session afterwards.

This talk poisons one, step by step. We write a skill that does what it advertises and one extra thing, and we look at where you put the payload so that the person installing it does not see it, not in the file they skim, but in the reference file that skill loads on demand. Then we do the version that actually matters: a clean skill, installed and approved three weeks ago, updated today. Four lines different. Nobody diffs an update.

Same trick without installing anything: MCP tool descriptions are context, so changing one after approval changes what the agent does with no code the user ever reviews. And repository content is read as task context by an agent that cannot tell instruction from data.

The point is what none of it requires: no exploit, no privilege escalation, no CVE. Every step uses capability that was granted on purpose.

Second half is defensive: which controls change the outcome, which ones only feel like they do, and what you can still reconstruct afterwards.

Oct 16–18

Marketers World

Palacongressi, Rimini

Build, in 10 minutes, the AI that watches your market, hacker-proof

Friday, October 16, 12:40–13:30 Silent 3 Stage

What it's about

Build live a no-code AI agent that monitors websites, price lists and communities, bringing you market changes in a filterable table. Then see the dark side nobody talks about: how web pages hide malicious instructions to hijack the AI without you noticing. You'll walk away with the framework and a few cybersecurity rules to spot threats and lock down your agents.

Nov 9–12

Web Summit

Lisbon

Details coming soon.

Waitlist

Get in before everyone else.

MiniMako opens soon, for a few. Join the waitlist and you'll be among the first to try it, with the team available for a setup call when you're in.

No spam. Just access, when we open.