Emma: A Private and Personable AI Assistant for Everyone

As an avid fan of generative AI and chatbots, I have always dreamed of creating my own personal assistant that could understand me and interact with me in a natural way. That's why I decided to embark on this ambitious project: Emerald Synthetic Intelligence (Emma), a full end-to-end AI assistant platform that runs on a local machine and respects your privacy. I will be using this blog to keep a development log for the project.

I plan on Emerald Synthetic Intelligence (Emma) to be a full end-to-end AI Assistant platform. Think ChatGPT meets Echo. The primary focus of the platform is privacy. First and foremost, I intend on it being capable of local hosting. From the embedded device through to the LLM, this system is intended to run on a local machine with modest capabilities. Second, I hope to create an AI assistant capable of displaying a personality of sorts. I want users to feel more like Emma is a friend than Siri or Alexa.

So, I think the big question I ask myself is why this project. Honestly, it is a brilliant combination of a number of my interests. I have been very involved in generative AI, and AI in general. I created my first Convolutional Neural Network project back in 2016. Since the release of ChatGPT my interest has only grown. Then there are the obvious Software Engineering challenges that I look forward to solving. And lastly, I get to work with hardware on this project. I find that to be very exciting. Though it has been a long time since I touched an embedded system (I think the TI MSP430 Launchpad was $4.30 and just becoming popular) I am very excited to get back into it.

I hope to update the dev log at least once a week. First post should be my thoughts on architecture of the system (or at least the first prototype).