I built SignalGraph with Arnau Ropero and Santiago Garcia during the {Tech: Europe} x Cala x OpenAI hackathon. In roughly 10 hours we developed a visual board where an agent proposes starting points and people explore properties, investments, roles, events and other relationships. Each step is preserved with its source, can be revisited, and reveals when two branches intersect. In one test we followed Aedas -> Neinor Homes -> Avenue Capital Group -> DELSEY -> DHI Company -> France. The project placed third.
Problem, stack and result
Problem solved
AI-assisted research application that turns an open question into an explorable graph. Instead of hiding reasoning behind a final answer, it shows each entity, relationship and source so people can choose which thread to follow.
Technologies used
Core stack: Next.js, React, TypeScript. Project technologies: Next.js, React, TypeScript, Cala AI, Knowledge Graphs, Data Visualization, LLMs.
What I did
My role was Co-developer · SignalGraph. I worked on the technical implementation, product approach and enough documentation for the result to be explained and evolved.
Result or learning
Third place at the {Tech: Europe} x Cala x OpenAI hackathon. Functional MVP built in roughly 10 hours with Arnau Ropero and Santiago Garcia.
Highlights
- Third place at the {Tech: Europe} x Cala x OpenAI hackathon.
- Functional MVP built in roughly 10 hours with Arnau Ropero and Santiago Garcia.
- Visual graph for following entities and relationships without losing research context.
- Every step preserves its source and supports branching, backtracking and evidence review.
- Cross-path detection turns hidden connections into inspectable findings.
- Test case: Aedas -> Neinor Homes -> Avenue Capital Group -> DELSEY -> DHI Company -> France.