==============
== Samswara ==
==============
Pursuing Agency

Giving Up Logjam

Why I’m giving up LogJam

I hate giving something up before it’s finished but:

  • I’m burning a ton of time on the UI and not even using it
  • I’m unconvinced productivity apps are worthwhile
  • It’s simply not as fun as paper and pen

Final Product

It’s a Mac OS X

LogJam Screenshot

What I got from it

  • Way more Rust exposure
  • How to bundle Mac apps (!!!)
  • The next thing

What does the ultimate productivity app look like?

The ultimate productivity tool is seamless.

It looks like this: You clip something on a pen. You write, it automatically transcribes into a cloud synced document.

Todo’s are marked with a special character, you can schedule reminders with natural language.

We can build this.

Here’s what we need:

  • Pen tracking sensor
  • Handwriting -> Text
  • Text -> (Instructions, Notes)

It’s easiest to approach in reverse order:

  1. LLM that parses a text file to a set of tasks with reminders
  2. A tool that takes a picture of handwriting and converts it to text in the system
  3. The pen tracking sensor (most difficult part)

But that doesn’t derisk the project well, since #3 is the hardest.

It actually makes sense to approach that first:

  • Pen tracking with an arbitrary pen

What does this project get me?

  • Hardware Experience
  • Machine Learning Experience
  • More Side Project

What are the risks?

  • Get’s frontrun by something like Tab, where you converse with the model
    • Might be easier to talk with a model than write
    • But something super unique about pen & paper…
    • Or is there? Maybe there’s something unique about talking?
    • Pen & Paper is very output focused, non distracting
    • There’s a reason humans use whiteboards while talking.

It’s worth a try, let’s do it - LogJam Too.