Simone Bova
He takes AI agents and company brains to production, and makes them hold when nobody is watching. He co-founded Yempik in Turin in 2025 with Raffaele Zarrelli and is its CTO.
Simone decides how the systems Yempik takes to production are built, and writes their code together with AI agents. He has one test: does the system hold real work, with real people, when nobody is watching?
That is why he puts the checks inside the code that performs each action, keeps the model replaceable and treats a person’s approval as part of the product. When something breaks, he writes down what happened and which rule came out of it.
He also works with Intarget on AI and AI product development, and builds WhatsApp, voice and email agents for startups and companies. He has run a workshop at Bocconi.
- Yempik Company BrainThe company memory that people and AI agents read and add to. Every new piece of information becomes official only after a person approves it. In production since late July 2026.
- Intarget caseThe system for gathering and consulting customer knowledge, built for Intarget’s Innovation Hub. Simone led it for Yempik.
- code-osThe open-source method Yempik uses to build software with AI agents: spec, plan and verification before every release. Simone maintains it.
- cowork-osOpen-source workspace (MIT licence) for Claude Cowork, published with Raffaele in late June 2026: decisions and context in files the team owns.
- Yempik Mail GatewayOpen-source server (MIT licence) that lets AI agents prepare and send email with drafts, sending limits and a way to stop everything. Maintained with Raffaele.
Simone writes in the first person on simonebova.com. These three pieces are about Yempik’s work and we republish them here with his consent. The original remains the reference.
- CONTEXT IS GOLD: a company brain the next agent can use without youContext is worth something when another agent can find it, tell when it holds, correct it and reuse it. Rules from Yempik's company brain and my second brain.
- Anyone can buy the model. Knowing which work to remove is rare.Anyone can use a frontier model. The edge is knowing which work to remove, which judgment to leave with the expert, and how to prove the result. Cases from Lia and CVing.
- Own the harness, swap the model. What you actually need to ownOwning the harness means owning state, authority, rules and verification, with native tools underneath. What I learned from DeepSWE, CVing and Yempik's Company Brain.
- simonebova.comAll his writing, in Italian, English and Polish.
Topics Simone speaks on at events, workshops and podcasts. Each one already has a published piece.
- Putting an AI agent into productionMethod matters more than the model: connected data, workflows designed for AI and the expert’s signature.
- Where AI value livesThe question about the impossible to ask at the first meeting, before any demo.
- Knowledge an agent can useWhen an approved piece of information is actually reliable, and how it passes from one agent to the next.
- What a company must own of its AIRules, checks and data stay with the company. The model can be swapped.
- Where knowledge dies: what a company loses when it works with AIJuly 12, 2026
- The impossible question: the AI value experts don’t seeJuly 11, 2026
- How to put an AI agent into productionJuly 2, 2026
- Why 95% of AI projects leave no markJuly 2, 2026
- From tacit knowledge to the agentJuly 2, 2026
- The Yempik case: how our Company Brain worksJuly 2, 2026
- What level is your company brain at?July 1, 2026
- How to build a company brain on filesJune 27, 2026
- Who owns your company brainJune 25, 2026
For interviews, talks, workshops or projects, write to simone@yempik.ai.