Ethan Vernet

The Tracy Server project is about laying the foundational central brick for all AI automation.

Before launching a pipeline, before deploying an agent, before automating anything, there needs to be a place where intelligence is centralized, prepared, and orchestrated.

This is exactly what Tracy Server does.

A single container deployed in your infrastructure. A secure API. Zero external dependencies.

Tracy Server centralizes agent preparation, AI model management, data structuring, and inter-agent communication in one place.

Everything your pipelines need to run is available in the same location, under your total control.

Result: controlled costs, eliminated dependencies, infrastructure that truly belongs to you.

Developed in Go, no Python, no no-code, no interpreted language. Performance and total control were not options, they were the number one priority.

What if your AI assistant ran entirely on your workstation, without ever sending a single piece of data outside?

This is the idea behind Tracy.

The future of artificial intelligence is local. Not in someone else's cloud. Not on servers you don't control.

On your machine. Under your total control.

Tracy is a conversational agent designed to run locally and ensure the absolute privacy of your data. It transforms your internal documents and information into a searchable knowledge base, without ever exposing your sensitive content to an external service.

Your data stays with you. Period.

Tracy combines RAG, native performance, and seamless integration into your existing setup to become your team's everyday research assistant.

Ask a question. Get an answer. From your own documents.

Locally. Confidentially.

What if our CPUs ran on light rather than electricity?

This is the dream that sparked this project. And it's exactly what we're exploring.

Funded by the National Research Agency, this project involves designing ALUs (Arithmetic Logic Units) by merging Artificial Intelligence and plasmonics.

A rare collaboration between three fields of expertise: physics, optics, and computer science. Three worlds, three languages, one vision: rethinking the foundations of computation.

Plasmonics allows light to be manipulated at the nanoscale. Where electricity hits the physical limits of silicon, light opens a new frontier — faster, more energy-efficient, denser.

The goal: to prove that it's possible to build the fundamental logic gates of computing by replacing electric current with light waves.