AI Research & Development
Our current focus for research and development is in the field of Artificial Intelligence and Machine Learning.
We have successfully applied AI to:
- Financial market forecasting (ENZO-TS)
- Virtual assistants and agents running on modern LLMs (ChatAI)
- Autopilot for airframes (XPSVR)
LLM-based Virtual Assistants and Agents
Function-calling and prompt-injection are used to give the assistant a sense of location and time of the user.
Parallel agents are used to fact-check the conversation and provide additional information on the subject being discussed. This approach, while requiring additional token budget, addresses the hallucination issue that arises from naive use of an LLM. The increase in reliability for the user is ultimately worth the additional cost, as it saves time that the user would otherwise have to spend for manual verification or, worse, risk being completely misinformed.
The assistant operates as a web app, leveraging Flask, Bootstrap, and additional external modules for code syntax highlighting and LaTeX rendering, but at the core it's a Python application that is adaptable to other platforms.
This is an ongoing project used both for deployment and internal research.
XPSVR (flight sim)
XPSVR is an experimental flight simulator built to mimic the F-35 Lighting II
jet fighter. The project targets VR for full-immersion.
Avionics display includes the famous 2-screen touch display of the F-35 and its windowing system that allows to configure the panels and to maximize and minimize various elements.
We employed machine learning for the autopilot and weapon systems. The targeting system uses basic ML to determine the launch envelope of ordnance, based on an internal simulation of the missile's flight that happens multiple times per second.
Rendering is performed with the OYK Game's rendering engine which we originally wrote for mobile games, and that we later extended for PC GPUs and OpenGL 4.5.
See more details on the XPSVR's blog.
ENZO Trading System
We've been involved with crypto trading since 2018, when we started the development of our own trading algorithmic trading system, ENZO-TS.
ENZO-TS is a complete system that allows for development of custom indicators, backtesting, and live trading on multiple exchanges. Trading client also include position sizing based on our proprietary risk management model, which has evolved greatly across the years.
ENZO-TS was build from the start for total automation of trading, leveraging machine learning. After much internal R&D, in 2023 we finally deployed our first algorithms based on Artificial Neural Networks (ANN).
For our model, we ultimately settled for a Deep Neural Network trained with a set of Evolution Strategies. This approach proved effective, particularly with limited computational resources. While backpropagation is generally more efficient and likely a better long-term solution, the evolutionary approach facilitated a better balance between exploration and exploitation.
See more at www.enzobot.com.
Our best-selling titles are Final Freeway, Finale Freeway 2R and Fractal Combat X, with several millions of downloads worldwide on App Store, Google Play, Amazon and other parallel distribution channels.
Media Translation Services
We are specialized in anime and manga translations from Japanese to Italian language. Our clients include Square Enix and Square Enix for which we've translated titles such as Fullmetal Alchemist, Naruto, and Black Lagoon.
For more details, please contact us.