Direction

Having a clear direction when you’re doing side projects is so important.

You have 30 minutes to do something? You want to be able to pick a task and start executing. Maybe the task isn’t the high leverage thing you can do. But it’s definetly valuable.

Having a direction requires something akin to product market fit, yet incredibly more basic. It requires momentum, getting over the hump.

For you to have direction you need to have built the beginnings of the engine. You need:

• A barely functioning prototype - you need to have targeted the main flow, so you can begin iterating on it.
• Passion - you need to believe in what you are building.
• A task board - you need a way to pass todo’s to your future self.
• Quick deploys - you want to build, merge, and deploy painlessly, so you can cognitively move on.

Basically, you need to be following a slope to a local maxima. This is completely separate from PMF. That comes much later, if you are lucky.

Let’s call this SPCF - Side Project Chris Fit. For potential side projects I need to be asking:

1. Can I build a working prototype in a few days to have a base for iteration?
2. Do I think it matters? Is it exciting?
3. Does it bring me closer to an edge of knowledge than I am now?

What SPCF doesn’t require:

• Long term potential - nothing that I do now is going to be the actual business I start. This is all about exploration. It doesn’t need to be practical.

Everything else follows from these two things. So here’s some potential ideas:

• Tunnel simulator - is somewhat fun but suffers from being too far removed from the actual product which requires hardware
• Tunnel robot - I could have this running pretty constantly delivering items in the house. This could be quite fun, but would need to be built!
• LLM research assistant - everyone’s doing it, but could I build a tool that helps you learn concepts faster?
  • I likely could. It will be outcompeted for sure, but it might be beautiful for a short period of time.
  • Would be UI heavy, but possibly in a pretty way.
  • Could use a local model with Mistral
  • Can solve real problems I have at work, and I could use it at work to dogfood it
    • What problems do I have at work? Wanting to bounce design ideas around, to visualize a design I’m working on, stuff like that.
• Better Splitwise
  • 0 pay gates
  • Reporting
  • Export to excel
  • No centralized backend:
    • IPFS
      • Probably cheap enough you can just subsidize it
    • Urbit
  • This is a very small amount of data, you could definetly find a fun way to do it cryptographically…
  • Target families
  • Definetly can start with just a backend
• LLM Meal Service - this is a real business, but could still be quite fun.
• AI Values Coach
  • You imbue the llm with values and then with RAG…already hate it
• An LLM powered todo list:
  • Automatically categorizes things
  • Automatically finds places in the day
  • Automatically outsources?
  • etc etc

I should keep brainstorming ideas with potential SPCF. There’s a lot of fun organizational things I want to do - setup a Notion, a Github, an LLC. All of those have to wait until we have SPCF.