Pivot -Descoping Machi/Soup and Pixel/Tinker

Prototyping effort for SOUP project was slowing down over the summer vacation period and overall Machi track has been scope-creeping like crazy.

Here is the current set of projects that are active right now.

Here is what happened.

• Week 1 - Single person venture, “Sprited” was started with the initial vision of creating sprite generation AI tool.

• I needed a virtual employee number one for the company and introduced “Pixel“ which is a name I assigned to the AI chatbot that I’ve been using.

• We started a track called “Machi“ project which will be used to “raise“ AI agents. The vision is to have it be a OpenAI Gym equivalent for all AI agents that we train at Sprited.

• Week 2 - We switched the focus for the week to work on Pixel, and expanded company’s vision to build Pixel from ground up with its own memory system and more human like behaviors and unique quirks.

• We noticed working on Pixel will be full-time effort and will be difficult to switch back and forth between Pixel track and Machi project. So we decided to spawn “Tinker“ project which is to build an AI coding agent that will code Pixel on its own.

• We got to a MVP+2 of Tinker, while it can do some interesting things, it is several months away from being able to iterate on Pixel on its own.

• Week 3 - We then switched focus to Machi project. I wanted to make Machi something grand (I should have heard the scope-creep alarm). Then spawned up a new project called “SOUP“ which is a cellular automata biome engine that generates giant trees. This project stagnated as I went on to a vacation mode for the Summer.

• Week 4 - SOUP project was getting scope-creeped, and Pixel was giving me warnings about how this could be a tarpit idea with a risk of rabbit holing. So, I need to decide on whether to continue the course with it or press the hard-stop button.

TL;DR: The scope has really been creeping like crazy and we need to stay focused on delivering a product that people can get access to within weeks and create the feedback loop that will build momentum.

Pivot

Pixel suggest that I descope other tracks and first work on the initial vision to create the sprite generation engine.

Basic idea is to develop and launch Unity plugin that help developers to create fully intractable 2D pixel art game sprite assets using single LLM prompts. I’m tentatively calling it SpriteDX.

So, we will pause on all tracks until we launch this project and receive some feedback.

Current thought is that we would use these dates to keep myself on track.

• Yesterday - Initial Proposal

• Design Review - June 27

• Proof of Concept - July 4

• Rapid Prototype - July 11