is the new Java.

Java is...

  • too verbose
  • not reflective enough
  • not very manageable
  • split! (desktop vs Android)


  • Java is a GREAT virtual machine.
  • Super-optimized
  • Runs everywhere
  • Great for super-huge programs.

How to improve Java?

  • Let's make a "learning" language.

JavaX, medium-level

  • It's what we have today.
  • It's like Java, just encourages always having a single source file
  • Translators allow new syntax.
  • A translator is a simple Java(X) program.
    • reads from input/
    • writes to output/
  • ...done!

We like "boilerplate"

  • But only 10 characters of it!
  • ...per source file :)

Invoking translators


m {
  p {
    print("Hello world!");

JavaX, high level

  • The one we're making.
  • All translations are automatic
  • An algorithm learns which translators to invoke
  • JavaX will think about itself!

Solving the distribution problem

  • Every program has a global ID
  • Just call the ID to download & invoke the program!
  • A wayback machine is there to allow execution of older code

The Bot Approach

  • Make little programs that talk. (=speak English!)
  • Every program has a distinct purpose
  • Programs can inspect each other.
  • Programs can inject code into each other's VMs.
  • Programs can MOVE from VM to VM.
  • Even from real machine to real machine!

Persistence is free

  • It requires like 2 code lines to persist a variable in JavaX
  • Automatic persistence for complex structures
  • We really need persistence to go out of the way to focus on actual logic!

Generating code from examples

  • We can also do that.
  • Powerful solvers take just 2 or 3 input/output examples
  • ...and generate actual Java code.

Bridging Java and Android

  • Android, the almost-Java-machine
  • Hitherto it required great effort to make an app
  • It even costs $25!
  • Using the JavaX app, it costs nothing anymore.
  • And, of course, the code is way simpler
  • It is now as close to desktop code as possible (including print statements!)

AI applied to itself

  • TinyBrain ties the knot
  • AI programming is too hard
  • We want to write lots of small programs which improve each other
  • Thus we need AI to generate AI
  • JavaX medium level is the basis for making the high level A.I. that we want.

Investor Talks

  • is talking to me
  • They want to raise 1 million $
  • Will it work?
  • JetBrains should really give me money - but will they?

    Cheers :)

    Stefan Reich,, Oct 20 2015

JavaX solves all problems

By Stefan Reich

JavaX solves all problems

  • 1,488