AI before AI — since 2008
Risk, portfolios, ALM, securities analysis: quantitative computations for banks, portfolio managers and financial institutions, built on twenty years of physics applied to markets. Today every client is served by Lisa, a dedicated AI agent: she answers, computes and delivers in the chat the client already uses. People, where they matter.
Meet Lisa
Let's talk data
Scroll
Today
Every client has their own Lisa
Since the summer of 2026, Redexe clients are served by dedicated AI agents. Lisa lives in the client's working group: she delivers the reports, answers questions on the numbers always quoting the analysis date, launches computations on the quantitative platform and posts progress updates. No app to install, no portal, no new password.
One agent per client, one perimeter per client: Lisa only sees the data of the client she serves, only talks to authorised contacts, and can only perform the operations approved for that client. When a request falls outside her perimeter she does not improvise: she logs it and hands it over to a person.
Within a few weeks we went from one agent to a fleet of agents in production. The founder now focuses on research and supervision: the routine — all of it — is Lisa's. We believe we are among the first firms in Italy to have taken this step all the way.
About
Science applied to data, since 2008
Redexe was founded in 2008 by Riccardo Donati, a physicist, as a spin-off of an asset management company. Its distinctive trait has never changed: models born in physics — complex systems, statistical mechanics, heavy-tailed distributions — applied to concrete problems of measurement, forecasting and decision-making.
The proprietary risk measure RedES™ has been published in peer-reviewed journals and tested on more than three thousand equities. Machine learning was in our production systems as early as 2020, before the boom of generative AI: "AI before AI" is not a slogan, it is a date you can verify in the trade press.
Today Redexe serves banks, portfolio managers, financial advisory firms and institutions in Italy and Switzerland, where it operates in partnership with Egonon SA. Redexe provides software and data-processing services, strictly B2B: investment decisions remain with the clients and their professionals. And the relationships last: our longest-standing clients have been with us for ten and twenty years.
Services
Managed services, not projects
Analyses in continuous production — daily, weekly, monthly, semi-annual — delivered by Lisa in the client's working channels. The common root: proprietary quantitative models, verified on data.
Ex-post reports and stress tests based on RedES™, the proprietary risk measure designed for shocks and bubbles: the model is declared first and judged after.
Read more →
Robomanagement™ in production: optimisation with AI-extracted sentiment from research documents, full NAV reconstruction, honest comparison against the no-sentiment baseline.
Read more →
Ten-year projection of assets and disbursements with a double balance-sheet Monte Carlo and human-validated commentary. In partnership with YouTrust & Partners, with the collaboration of Ca' Foscari University of Venice.
Read more →
RedRating updated every market day and bond analysis with human oversight: scenarios, probabilities and declared disproof triggers, with a longitudinal memory per security.
Read more →
Critical infrastructures for a few selected clients, with the reliability standards of finance and relationships that last decades.
Read more →
In partnership with LoveImmobiliare: artificial intelligence applied to property valuation and to the agency's daily operations.
Read more →
Method
The method: physics, AI and people
Tails first. Our risk measures work on Pareto-Lévy stable statistics, not on the normal distribution: they capture extreme events exactly where traditional measures fail. And the model is declared first — parameters fixed ex-ante — to be judged after, on realised results.
Artificial intelligence where it helps, with discipline: structured research by scenarios with probabilities and disproof triggers verified at every run, sentiment extracted from documents, commentary in plain language. And humans where it matters: oversight of sensitive content, declared validation and escalation. AI-generated texts carry their transparency note.
Underneath, an industrial discipline: inputs are validated before any computation and errors are explained in plain language; every run is tracked and reproducible; every release is verified number by number against history. Not a tool: a production line. Decisions are made on numbers — and the same rule applies to this website, maintained by AI agents under human direction.
The numbers
Facts, not promises
All verifiable, none of them marketing:
Research & heritage
Where we come from
The activities that built the method. They are no longer active businesses: what remains are dates, publications and verifiable results.
FAQ
Questions & answers
What does Redexe do?
It provides software and quantitative data-processing services to banks, portfolio managers, financial advisory firms and institutions: risk measurement, portfolio-construction computations, ALM for foundations, model-based analysis of equities and bonds. Results are delivered as ready-to-use reports in the client's working channels.
Who is Lisa?
Redexe's AI agent, in a dedicated instance for each client: she answers questions on the client's data, launches computations on the quantitative platform, delivers reports and simulations in the working group's chat. She is an AI framework developed by Redexe, with human supervision and separate perimeters per client.
Are Redexe's analyses investment advice?
No. Redexe does not provide investment advice: it supplies software and data-processing services to institutional clients. The analyses are model-based computations supporting decisions; investment choices remain with the clients and their licensed professionals. Nothing on this website is a recommendation or a solicitation to invest.
Is client data safe?
Each client has a separate perimeter: their agent only sees their data, only talks to authorised contacts and can only perform approved operations. The computation method remains Redexe's; the results belong to the client. Irreversible actions require explicit human confirmation.
Who does Redexe work for?
Strictly B2B: banks, portfolio managers, financial advisory firms, foundations and institutions in Italy and Switzerland, where Redexe operates in partnership with Egonon SA and with YouTrust & Partners SA (financial advisory, Lugano). No retail services and no personal financial advice.
What technology does Redexe use?
A proprietary quantitative platform developed over twenty years — with models such as RedES™ (risk measure), Robomanagement™ (virtualised management team) and RedRating (equity rating) — plus a state-of-the-art artificial-intelligence layer for research, commentary and agents. Lisa is an AI framework developed by Redexe.
How long has Redexe existed?
Since 2008. It was founded by Riccardo Donati, a physicist, as a spin-off of an asset management company. It was running machine learning in production as early as 2020, before the era of generative AI.
How can I contact Redexe?
On LinkedIn, by messaging Riccardo Donati: that is the website's contact channel.