For a startup project led by Kevin Taulera, I was hired to develop a web application called Supply Finder.
The goal of this application was to assist media traders by simplifying the process of creating programmatic advertising deals across multiple platforms.
With Supply Finder, a media trader can enter a brief in natural language — for example: “France, display, target audience: house owners, 30–50yo” — and instantly receive deal recommendations compatible with several major supply platforms such as Equativ, Index, and Xandr.
Once the recommendations are generated, the user can easily create and manage deals simultaneously across these platforms, significantly reducing the manual effort and complexity usually involved in the process.
This project combined natural language processing for parsing the brief, integration with multiple supply-side platforms’ APIs, and a streamlined React-based interface to provide media traders with a fast, intuitive, and scalable workflow.
https://www.youtube.com/watch?v=FhhwROQr3WI
Processing a brief
The main feature of this software is the processing of a brief: it allows a user to enter a deal request in natural language and generate multiple deal proposals across different supply platforms. The system I designed parses the brief for values such as format, target audience, publishers, and more, then maps these values to the corresponding parameters in various supply platforms (Equativ, Index, Xandr) using a complex RAG system, in order to suggest compatible deals.
On this same menu, users can select the deals they want to keep and create, and the system will then publish them in the corresponding supply platforms.
Modifying a deal
This popup allows the user to manually modify a deal, even after its creation.
Modifying a seat
This menu allows the user to set their seat IDs and DSPs.
List briefs
This menu allows users to view their past briefs and edit or delete them. Briefs are shared across the entire organization the user belongs to.