About · Made in Italy

One AI engine of our own. Two products built on it.

We design the Vivi engines and put them to work where the data is: on your server and inside your devices.

Start free → Contact
2013
Together since
3
Products, one engine
<20KB
Model on an MCU
2
bioRxiv preprints
§01The engine

One engine, two products.

Vivi is our family of AI engines. It learns on its own what normal looks like in your data.

  1. 01 Shield Website protection Stops bots, scrapers and traffic waves. Customers and Google get through. →
  2. 02 Lynx Server monitoring Notices on its own when a server or container misbehaves. →
§02Principles

Four rules we don't bend.

01

On the device

Decisions happen where the data is. No latency, no mandatory cloud.

02

Your AI stays yours

Models run only on your hardware, bound to the chip's ID.

03

Simple

Upload a CSV, download the binary. For servers, one Docker command.

04

No lock-in

No ML expertise needed. No cloud dependency.

§03Team

Two founders, together since 2013.

From Derved, a design studio, to Docfire, to Luviner.

Co-Founder & Strategy

Jessica Ferrari

Engineer, PhD in Applied Acoustics. Leads strategy, clients and business development. Part of every project since 2013.

Co-Founder & CTO

Filippo Groppi

Full-stack developer (Python, PHP, C), founder of Docfire. Built Luviner's AI engines, the platform and 399 automated tests from scratch, with no outside funding.

§04Research

Our research is public.

bioRxiv preprints, authored by both founders.

  1. [1] Groppi F., Ferrari J. (2026). The same plasticity rule rescues networks built from one conductance set and destabilises networks built from another. bioRxiv. doi:10.64898/2026.09.15.751713
  2. [2] Groppi F., Ferrari J. (2026). Weak pre-before-post pair biases sort synaptic weights in a conductance-based model of spontaneous network bursting. bioRxiv. doi:10.64898/2026.09.03.749248

Try the engine on your data.

Free to start. No card required.

Start free →

Investors

Proprietary engines, no GPU clusters for inference. Revenue grows with customer devices, not with compute.

Get in touch →