Context
The emergence of large language models created a unique opportunity to rethink how investors interact with their data. While Edda already managed millions of data points for investment funds across the globe, artificial intelligence promised to transform this wealth of information into actionable insights and measurable productivity gains.
The challenge wasn't simply to integrate an AI API into the product, but to design an experience that would make artificial intelligence natural, reliable, and genuinely useful in the demanding context of investment management.
Problem & objectives
Investment teams spent hours on repetitive tasks: writing deal memos, analyzing financial documents, extracting key information from dozens of PDFs. These manual processes created bottlenecks and delayed critical investment decisions.
Our objectives were threefold:
Drastically reduce time spent on low-value tasks while maintaining the accuracy and rigor expected in finance
Create an AI experience that integrates naturally into existing workflows without requiring extensive training
Guarantee reliability and transparency of AI-generated results to maintain trust among professional users
Solution description
Intelligent content generation
I designed an AI-assisted generation system that transforms structured deal data into coherent narratives. Users can instantly generate a complete deal memo, one-pager, or executive summary by simply selecting the desired format.
The key to this feature lies in its contextual prompt system that understands investment data structure and adapts tone and detail level according to document type. The AI invents nothing; it structures and articulates information already present in the system.
Document extraction & analysis
For complex financial documents, I developed an intelligent extraction system that automatically identifies and structures key information: financial metrics, deal terms, team bios, use of funds plans. Users simply drop a pitch deck or financial report, and AI automatically populates relevant deal fields.
The interface clearly displays the source of each extracted piece of information, allowing users to verify and adjust with a single click. This transparency was essential to gain trust from users working with tens of millions of euros.
Review & control interface
Knowing that AI must remain an assistant and not a decision-maker, I created an experience that firmly places humans in control. Each generated content appears in an edit mode where users can refine, correct, or regenerate with specific instructions.
The system keeps version history and allows easy switching between original content and AI-enhanced content, giving users a complete sense of control over their data.
Monetization & experimentation
Introduced a credit-based usage model to test willingness to pay and validate value perception.
The monetization framework remains evolvable, but provided critical insights into user behavior and AI feature adoption.
Product learning & data structuring
The AI rollout served as a structured test-and-learn phase.
Beyond feature delivery, it helped:
Improve data normalization and structuring
Identify scalable AI use cases
Strengthen product experimentation capabilities
Positioning AI as a product capability, not a gimmick.
Other projects/
Want to work together? /
I’m currently available for new collaborations; short or mid-term projects, full-time roles, or advisory work.
From product strategy to hands-on design and execution, I support teams across the entire product lifecycle.
If it sounds relevant, let’s set up a 30-minute call to explore fit.
