The publisher of Osmoze

Heavenize

Publisher of the Osmoze cockpit for the Front Office of asset management firms.
At the intersection of the business, AI and software publishing.

Our story

20 years of R&D: engine in production since 2012, AI layer recently delivered in the product.
2006
Start of design
(individual R&D)
2008
Heavenize
founded
2012
Patent FR 2.994.295
+ entry into production
2019
Osmoze 3.0
Front Office refocus
2024
Model-agnostic
LLM layer
2025
Osmoze 5.0: AI
at the heart of the Front Office
2026
Agentic layer
+ Portfolio Autopilot R&D

An architecture designed
for intent

Why AI is not a 2024 pivot at Heavenize, but the result of an early insight.

It all starts in 2006, in individual R&D, from a technical conviction: capture the portfolio manager's intent, rather than impose a standard workflow on them. Heavenize is founded in 2008 to bring the concept to maturity. The patent is filed by Heavenize in 2012 (FR 2.994.295 , "Method for creating a software application implemented by an engine"), at the time of entry into production at the first clients. Nearly twenty years later, AI arrives with exactly the same problem: interpreting an intent expressed in natural language, and executing it with business-grade rigour.

The Intentional Engine was ready before AI. That is why the two live in the same architecture here, not out of opportunism, but by design.

The convergence is documented

Thirteen years before modern AI agents, the orchestration pattern had already been formalised by Heavenize.

In 2012, we formalised an approach that closely resembles what the market today calls agentic AI: a goal-oriented engine, able to select, execute and compose dynamically modules to produce a result. Since then, this orchestration engine has been industrialised and runs in production in capital markets: first to orchestrate deterministic business modules, and now as the foundation of an AI layer recently delivered that brings LLMs into the same architecture.

Patent FR 2.994.295 , filed by Heavenize, describes a precise mechanism: an engine receives an objective, queries a base of interpretable objects (each accompanied by the description of what it returns), executes the relevant module, compiles the results, and replans as long as the objective is not reached.

This is exactly what structures modern AI agents today, under the names function calling, Model Context Protocol (MCP), ReAct. When the industry rediscovers this agent / tools / replan orchestration pattern in 2024, our engine has already been running it since 2012, then on deterministic business objects. The AI layer that brings LLMs into the same architecture has just been delivered in the product.

Filed in 2012, this patent stands first as dated, publicly verifiable proof (INPI) of the prior conception of this design: an official timestamp, thirteen years before the industry gave the pattern a name.

Same mechanics. Thirteen years apart.
Not a passing trend, but a convergence of design.

Our mission
Provide asset management firms with Front Office tools that embed their differentiating know-how,
rather than forcing them to conform to a market standard.

Why Heavenize,
not someone else?

We hold a positioning few competitors combine

The Front Office, seen from the inside

Twenty years serving portfolio managers, first internally at Natixis AM, AXA-IM, SG-CIB, then as a publisher. We are not a PMS publisher adding a Front module, nor an IT consultant discovering portfolio management.

AI, mastered in depth

A team that speaks both the language of portfolio managers and that of data scientists. Specialised agents, fine-tuning, RAG, multi-LLM: we build the bricks, not the marketing veneer.

Software publishing as a craft

A real server engine, an Excel add-in, open APIs: a technical foundation industrialised over 15 years, not a prototype dressed up as a product.

The team

José Chillan

President & Founder

A graduate of École polytechnique (X89) and a PhD in applied mathematics (M2 thesis on neural networks, 1993), José began his career at Dassault Systèmes, then at Marvin Software (Thomson-Reuters) as a quantitative analyst. From 2001 onwards, he worked alongside the Front Office teams of Natixis AM, AXA-IM and SG-CIB.

In 2008, he founded Heavenize to address at the source the problems he saw repeating across every assignment: PMSs ill-suited to the Front, Excel shadow IT, underestimated operational risk.

LinkedIn

Panongbene Sawadogo

AI Expert

A graduate of École polytechnique (X2017) and holder of two M2 degrees in Data Science and in Machine Learning, Panongbene leads the development of Osmoze's AI agents: multi-LLM architectures, fine-tuning on internal data, RAG and MCP integration, model selection and control of inference costs.

LinkedIn

Charlie Champanhet

Architect

A graduate of INSA Lyon, with 10 years of experience in multi-technology development, Charlie designs the technical architecture of the Osmoze cockpit: real-time server engine, Excel add-in, APIs, integration with PMSs and data providers. He guarantees the robustness and scalability of the foundation on which all business modules and AI agents rely.

LinkedIn

Rodolphe Aymard

Technical Expert

A graduate of École polytechnique (X89) and Télécom Paris (ENST), Rodolphe began his career at Dassault Systèmes, then, as a partner, helped build the company Itelios (an IT consultancy specialised in e-commerce, acquired by Capgemini in 2017). He works on structural technical choices: architecture, data visualisation, integration and industrialisation.

LinkedIn

Advisors

Two experienced executives from the software and financial ecosystem who support Heavenize on its structural choices.

Emmanuel Gillet

Advisor

A graduate of Télécom Paris. Today President of Powens and Evollis. Former CEO of Cassiopae, sold to Sopra Banking in 2017. Brings his long-term vision as a B2B fintech software operator.

LinkedIn

Alain Malochet

Advisor

Thirty years in the financial ecosystem. Co-founder of Dafsa, Diagram, Smart Trade Technologies and Omega FS. Brings his expertise on financial modelling and structured products.

LinkedIn

Asset management firms and insurance companies that have deployed Osmoze

Asset Manager

Natixis Investment
Managers International

International asset manager (BPCE)

Insurance

CNP Assurances

Insurance / Institutional investment

Insurance

AFI-ESCA

Insurance company · Insurance Front Office

Asset Manager

Apicil
Asset Management

Insurance-affiliated asset manager

Asset Manager

DOM Finance

Entrepreneurial asset management firm

Technology partners

CACEIS
Microsoft Partner

Long-term sustainability and business continuity

Continuous production since 2012, and the guarantees that come with it.

Osmoze has been in production since 2012 at French asset management firms and institutional insurance companies. Fourteen years of continuous use, without major interruption, on critical scopes (Front Office, compliance, order placement). The engine's maturity is not a promise, it is an operational track record.

Source code in escrow

The source code of the Intentional Engine is deposited with a trusted third party. Contractual trigger conditions: cessation of Heavenize's activity, hostile takeover, prolonged failure of support. Documented reversibility: every client can retrieve their data, settings and calculations in an open format.

Cloud or on-premise

SaaS hosting on Azure or AWS, or on-premise installation inside your infrastructure when the security context requires it. The infrastructure choice stays in the client's hands.

Support commitment

Contractual SLA adapted to Front Office criticality. Exhaustive technical documentation, knowledge transfer possible to an internal team or a third-party integrator if needed.

Want to get to know us?

The best way to understand Heavenize is to talk with us. 30 minutes between practitioners, no slides, no sales pitch.

Book a meeting