Two Years, One AI: Introducing IRIS, My Custom-Built PC Voice Assistant
I've always been fascinated by the promise of AI assistants, but I grew tired of the limitations of closed ecosystems. I wanted something more personal, powerful, and private—an AI that ran on my own terms, on my own PC. Two years ago, I started an ambitious project to build it. Today, I want to introduce you to IRIS.
IRIS is a custom voice chatbot, written entirely in Python, designed to be a clean, useful, and deeply integrated assistant for home PC use. It wasn't built overnight. It was born from two years of meticulous work, prompting out code, debugging, and refining a vision for what a truly personal AI could be.
The Core Philosophy: Local, Private, and Unrestricted
The most important design choice for IRIS was to run everything locally. Instead of relying on cloud APIs from big tech companies, IRIS is powered by Ollama, allowing it to run a variety of open-source language models directly on my machine. This provides two massive advantages: complete privacy and unrestricted conversations. IRIS can have deep, nuanced discussions without the content filters or data monitoring that come with commercial services.
The Tech Stack Under the Hood
For those interested in the nuts and bolts, here’s a look at the core components that bring IRIS to life:
Programming Language: 100% Python, making it versatile and robust.
The "Brain": A suite of local language models managed by Ollama. This allows for swapping different models depending on the task and ensures the core logic is private.
Voice I/O: The entire experience is conversational, using advanced Speech-to-Text (STT) and Text-to-Speech (TTS) libraries to handle voice commands and spoken responses.
GUI: IRIS features a sleek, "Glass-like" transparent interface that overlays on the desktop, providing visual feedback without being intrusive.
What Can IRIS Actually Do?
This project was all about creating a genuinely useful tool. Here are some of the key features I spent the last two years perfecting:
Real-Time Internet Research: I can ask IRIS a question, and it will browse the web in real-time to find, synthesize, and deliver the answer. It’s like having a research assistant who can summarize articles, find facts, and report back in seconds.
On-Demand Image Generation: Need a visual? I can simply ask, "IRIS, create an image of a futuristic city at sunset," and it will generate a high-quality image right on my desktop.
Deep, Unrestricted Conversations: Thanks to the uncensored nature of local models, IRIS can explore complex, philosophical, or creative topics without hitting a wall. This makes it a powerful partner for brainstorming and ideation.
A True Desktop Companion: From managing files and opening applications to having a genuine conversation, IRIS is designed to be a seamless part of the home PC experience.
The Journey: Two Years of AI Collaboration
Building IRIS was a marathon, not a sprint. The entire process was a unique collaboration between me and other AI models. I spent countless hours prompting, generating code snippets, debugging, and stitching them together into a coherent, functional application. It was a testament to how a dedicated person with a clear vision can leverage AI as a tool to build something truly complex and new.
This is just the beginning for IRIS, and it’s the perfect representation of what’s next for this blog. I’m excited to share more tutorials, project updates, and insights into the world of local, practical AI.
