“Alice had begun to think that very few things were truly impossible”. (Alice’s Adventures in Wonderland – Lewis Carroll)

In this article we retrace some events, taken from interviews by Piero Savastano, that led to the birth of the Cheshire Cat AI project…

• How did the name “Cheshire Cat” come about?
  Initially, Piero Savastano (the creator of the project) was torn between two names: “Cheshire Cat” and “Pinocchio”. Pinocchio was a fascinating choice for its connection to Italian culture and for the metaphor it embodies: a puppet who aspires to become a real boy, but who gets lost in a world of lies and illusions. A perfectly fitting figure, according to Piero, for an artificial intelligence project. The other option, Cheshire Cat, came from Piero’s passion for Lewis Carroll’s original book, Alice in Wonderland. In particular, he was struck by the scene in which the Queen of Hearts wants to behead the Cheshire Cat for his inability to beat him at cricket, but the Cheshire Cat responds: “Good! Cut off my head, I only have my head…”, alluding to the fact that the body had already disappeared. This cryptic way of speaking, from an impartial connoisseur who seems to know everything but doesn’t reveal it, was an intriguing image for an AI project. In the end, the name Cheshire Cat was chosen on the advice of a designer friend of Piero, who suggested it to him via Instagram…

• Who is Piero Savastano, the person behind the Cheshire Cat project?
  Piero started out as a researcher, then moved on to private consultancy and training in areas such as Machine Learning, Data Science and Deep Learning. As a boy, at 19, even before university, he began to deal with neural networks. He was fascinated by the human mind and its infinite nuances, but he also wanted a solid scientific basis: he wondered, looking at diagrams of neurons on some websites of the time, if it was possible to simulate a brain in a computer. This is where his vocation was born. He studied experimental cognitive psychology, with a focus on neuroscience rather than clinical psychology, where he worked with simulations of nervous tissue. Later, he collaborated with the CNR as a researcher, dedicating himself to artificial life simulations, a discipline more advanced and ambitious than AI itself, which simulates not only the functioning of the nervous tissue of organisms, but also the evolutionary processes of the species. After this experience, he worked as a freelancer and dedicated himself to dissemination on platforms such as YouTube and TikTok.

• When was Cheshire Cat born?
  During the ChatGPT explosion between November and December 2022, Piero was renovating his house. Having already worked with GPT-3 and created chatbots, he understood the impact of these tools and started talking about them on YouTube and TikTok, arguing that ChatGPT was just the beginning of a revolution: “These objects are just the beginning because they should be seen as language modules within more complex programs…”. To explain better, Piero published a tutorial in Python on GitHub, where he showed how to use these objects for more advanced applications. Starting from that tutorial and first public commit, the community grew rapidly, with forks and “stars” that increased rapidly. Between February and March 2023, Cheshire Cat had become one of the most used Italian open source frameworks for creating assistants based on language models. Now, it has thousands of downloads and an active community on Discord, with contributors from various countries. At the heart of the project are many Italians, including young talents and professionals with experience in open source and AI. The “core team” is made up of about ten people, while more than forty have contributed to the code, and there are hundreds of active developers who use the Cheshire Cat to create installations.

• What is the Cheshire Cat?
  The Cheshire Cat can be described like this: imagine you want to develop an application in which you want to integrate AI services. The goal is to work both at a low level (to customize the use of models and have the maximum possible control) and at a higher level, to avoid reinventing the wheel and rewriting everything from scratch. This is where the Cheshire Cat comes in. Basic Language Models receive a “prompt” and return a response, but the Cheshire Cat allows you to create dynamic prompts, enriched with context based on user interaction. In this sense, it works as a superior platform to LangChain: it uses LangChain, vector databases, Docker for networking, and provides a preconfigured microservice, integrable into existing architectures. So, rather than being a simple library, the Cheshire Cat is a configurable microservice for managing LLM and AI, which easily integrates with applications via websocket. It also provides plugins and customization primitives that allow you to build flexible AI applications without having to start from scratch every time.