Ardian and Mistral AI

1. 2X increase in weekly active users 
2. 2X growth in documents processed
3. 200K+ AI-powered queries generated

As a world leading private investment firm, Ardian is managing $192 billion in assets across private equity, real assets and credit with 1,100 employees and 20 offices worldwide. Ardian teams process vast volumes of unstructured documents daily—from market and due diligence reports to investment memoranda. To accelerate deal execution and enhance decision-making capabilities, Ardian partnered with Artefact and Mistral AI to develop GAIA, an internal Generative AI platform that transforms how investment teams analyze and extract insights from complex documents.

For the first time, enterprise-grade tech worked well with unstructured data. The ability to process documents quickly and accurately is essential. Generative AI helps teams rapidly understand documents and execute faster.

Arthur Garnier, Senior Data Scientist and IT Chief of Staff at Ardian

Document overload and information silos are universal challenges in private equity.

Private equity has evolved into an increasingly competitive and data-intensive industry where speed and accuracy are paramount factors of success. Ardian's investment teams  need to screen through hundreds of pages of due diligence documents, financial reports, and legal filings for each potential deal as efficiently as possible and extract the right information to support their analysis and decision-making. While human analysis is still key, empowering teams with the ability to extract this information more swiftly is now crucial to stay ahead of the market. 

"Teams receive long, complex documents—hundreds of pages. It's critical to distinguish signal from noise quickly," Arthur explained. The firm faced two primary challenges: accessing internal documents scattered across various systems under strict NDA requirements, and the sheer volume of information that needed rapid analysis. It was also critical to avoid "Information overload" that could prevent teams from focusing on high-value strategic analysis.

Before implementing generative AI solutions, Ardian’s data science approach focused mainly on numerical data and traditional machine learning models. While these methods continue to deliver strong value, the nature of private equity work—heavily reliant on unstructured text, PDFs, and qualitative analysis—meant they also needed to address unstructured content. GenAI now enables them to take this to the next level, allowing users to directly analyze unstructured data in their day-to-day operations.

Enabling Ardian's document intelligence transformation.

Recognizing the transformative potential of generative AI, Ardian developed GAIA, an internal AI platform built in partnership with Artefact and powered by a selection of AI models including Mistral AI. The platform launched in full production within one year, quickly becoming an essential tool for investment professionals across the firm.

"We realized we couldn't rely solely on OpenAI. We needed model flexibility—so we added Mistral models, which turned out to be very efficient, especially with French documents and summarization tasks," Arthur noted. The platform's model-agnostic approach allows users to choose between different AI models optimized for specific tasks, with Mistral AI excelling particularly in summarization and processing French-language documents—crucial for a European-focused firm.  

Beyond investment teams, GAIA enables all teams to upload internal documents and interact with them through natural language queries, providing rapid first-pass analysis of complex materials. The platform includes web search capabilities and allows users to create permanent folders for team collaboration, all while maintaining strict security protocols and data governance essential for handling confidential financial information.

The technical architecture prioritizes security and compliance, with manual document uploads providing an additional layer of human oversight

Why Mistral AI?

Ardian selected Mistral AI models for their:

• European data sovereignty alignment with private equity compliance requirements;
• superior multilingual performance, particularly with French documents critical for European deals;
• excellent summarization capabilities for processing large financial document corpora;
• model flexibility enabling specialized optimization for different financial analysis tasks.

Dramatic acceleration in deal analysis and enhanced team productivity.

The adoption of GAIA has exceeded expectations, with over 500 weekly active users. The platform has processed 18,000+ documents since inception and generated more than 280,000 user requests, demonstrating strong internal momentum and practical value delivery.

"The main difference now is that teams have much more clarified and much more organized information. It allows extra room for higher value-added tasks," Arthur observed. Rather than simply reducing time spent on analysis, GAIA enables teams to conduct more thorough due diligence and potentially evaluate additional opportunities.

Looking ahead: Expanding beyond document analysis.

Building on GAIA's success, Ardian is enhancing the platform with advanced reasoning capabilities for complex deal evaluation. The next phase includes automated subquery generation using Mistral AI's reasoning capabilities, enabling users to ask broad questions and receive comprehensive analysis across multiple financial risk categories.

"I would love to have a user asking a very broad, very general question, like what are the risks of this deal, and then having a proper retrieval model that understands what we are interested in as a private equity firm," explains Garnier, highlighting plans for increasingly sophisticated AI-powered investment analysis that understands private equity-specific financial metrics and risk frameworks.