Experts at the center.
AI agents don’t improve themselves. Experts do. We design every part of Syllotips around the people who know the work, so their time goes where it matters and nothing they teach is lost.
About SyllotipsAI built by researchers.
AI built by researchers.
Taught by experts.
Most enterprise AI is deployed once and left to guess. When it gets something wrong, someone patches a prompt, and a few weeks later the same mistake comes back.
Syllotips works differently. We don’t build another model. We build the continuous learning & improvement layer that sits on top of the AI agents you already use, grounded in research on how agents behave, fail and should be evaluated.
That research moves us forward. But since day one, we’ve known that what makes AI agents truly reliable is expertise. Expertise is more than knowledge. Knowledge gets locked in manuals and goes stale fast. Expertise is the awareness of exceptions, the experience behind every workaround, the judgment of knowing when a rule applies and when it doesn’t. It lives with your experts, not in your documents.
We build the system. Your experts teach it. It keeps everything they teach.
That’s our Closed Loop.
How it started
We founded Syllotips in Rome in 2024, after going through Techstars. The idea had come earlier, from something we had seen up close in our years in academia and in business: in every organisation, the most valuable know-how is rarely written down. It’s carried by people, passed on by the coffee machine, and too often leaves when they do.
Our name comes from that idea: syl-, from the ancient Greek for “together”, and tips. Expertise, shared.
When AI agents reached the enterprise the problem got sharper. Agents started answering at scale from incomplete, ungoverned documentation, and whatever the documentation left out, every answer left out too. The gaps weren’t in the models. They were in everything the models had never been taught.
So we built the Continuous Improvement Layer that closes them. When an agent falls short, Syllotips detects it and routes the case to the right expert. The expert reviews it, and what they validate is written back into governed memory, where every agent can reuse it.
That’s how the loop closes: agents learn what no book holds.
Research follows the same logic. It starts where the literature ends, and that’s where our team works.
Built by researchers.
Our R&D team is made up of several researchers, engineers and applied AI scientists, led by our Head of AI, Andrea Rossi. Together, they work on the questions that decide whether an agent can be trusted in production: how to evaluate it, and how to make it learn from real work. Testing the model alone is no longer enough. What matters is the whole system: its tools, its context, its memory, and its ability to get better over time.
We work closely with universities including Sapienza University of Rome and UC Berkeley, and part of that work is doctoral research we fund on how to measure the reliability of AI agents in real-world settings. We test what we learn where it counts: on real work, with real experts.
Today our team spans Europe and the US, with offices in Rome, London and New York. We work with enterprises in telco, insurance and energy, including EOLO and Leonardo Assicurazioni.
Our backers include Exor Ventures, Techstars, Azimut, Leo Capital. Along the way, we joined Harvard Innovation Labs and reached the finals of the Harvard PIC Challenge.
Security is part of the product, not an add-on. Syllotips is SOC 2 Type II and ISO 9001 certified, ISO 27001, and GDPR compliant. Its SaaS service holds the ACN cloud qualification at level QC1.
What we stand for
Who we build around, how far we push, and what we are willing to claim.
Push the edge.
We go after the problems the field hasn’t solved yet, not the ones that are easy to repackage. If a question is still open, that’s where we want to be.
Proof over promise.
Everything we ship has to hold up in production. We test, measure and question the result before we claim it. When we share a number, we share its scope. If we haven’t proven it, we don’t sell it.
A future where AI and expertise make each other better.
AI brings speed, scale and memory, while experts bring judgment, context and experience. Neither works as well alone.
To make enterprise AI agents reliable by keeping experts in the loop.
We turn every correction they make into governed, auditable knowledge your whole organisation can build on. What an expert fixes once stays fixed.
Founders
Two AI researchers and an engineering leader.
Giorgio Barnabò, PhD
Co-Founder & CEOGiorgio holds a PhD in Data Science from Sapienza and Pompeu Fabra University in Barcelona, with research on NLP and graph neural networks for misinformation detection and on algorithmic fairness. He has also worked on Alexa AI at Amazon. The first spark for Syllotips came at Sapienza: in a university full of experts, no one had mapped who knew what. At Syllotips, he leads the company’s strategy and its growth in the US, and spends most of his time with CIOs and AI leaders, helping them take AI agents from pilot to production.
Leonardo Martini, PhD
Co-Founder & CTOLeonardo holds a PhD in Data Science from Sapienza, where he met Giorgio, and degrees in Computer Engineering and Mathematics, both summa cum laude. During his PhD, he was a research intern at the Network Medicine Division of Harvard Medical School. At Syllotips, he leads the software’s architecture, from agent orchestration to governed memory, with one rule: make sure the structure holds before adding features.
Simone Silvestri
Co-Founder & CPOSimone studied Computer Engineering at Sapienza and started his career at Ericsson, working on 5G core networks and security, where he went on to lead a customer-facing security automation team. He joined Syllotips to build its back end and grew with the company into product leadership. Today he leads the product with an engineer’s rigor and a user’s focus: AI agents reliable enough for the enterprise, and simple enough for every expert to use, not just developers.
Board
Leaders who have built, bought and scaled enterprise technology.
Vincenzo Esposito
CEO of Microsoft Italy since 2023, Vincenzo brings the view from inside the ecosystem Syllotips is built to work with, and from the companies running on it.
Nicoletta Mastropietro
Nicoletta has led technology transformation as CIO of Leonardo, Vodafone Italia and Vodafone Group, and as Head of Innovation, Digital and R&D at A2A. She brings the perspective of the person who signs off on enterprise AI.
Maurizio Bobbio
Maurizio spent over twenty years at Oracle, most recently as Senior Vice President, leading Professional Services, Consulting and Customer Success across EMEA and Asia-Pacific, with around 7,000 people in 35 countries. Today he advises and invests in AI and security startups.
Lorenzo Congiu
Lorenzo is an Investment Manager at Azimut Libera Impresa SGR and has sat on the Syllotips board since 2025. A former startup founder with about ten years in tech, he now works with the boards of several early-stage companies and is on the faculty of H-FARM College. He brings both the founder’s and the investor’s view: what it takes to turn a strong product into a company that scales.
Your next failure.
Your next improvement.
Bring a real agent or workflow. We’ll show you where Syllotips fits, what it can improve, and how to get started.