Playing with Krobar.ai

Mike's Notes

For some years now, I have subscribed to the Kromatic weekly email newsletter. It is one of the best. This is no innovation theatre, which is refreshing compared to NZ. I really like using mathematical analysis to test ideas. Especially when Tristan explains how to do it.

It was here that I first learned how to use Monte Carlo Standard Deviation analysis using an Excel spreadsheet designed by Tristan Kromer. I then incorporated Monte Carlo into Pipi 9 to deal with uncertainty.

I had actually been using Monte Carlo in Pipi 6 for many years since 2017, but I did not know it was called that. It was buried deep in algorithms inside an early module that later became an engine.

Steve Blank, Alexander Osterwalder, Jason Cohen and Tristan Kromer are all top-notch. No bullshit. Everyone else is just a clone.

I had a free office hour with Tristan Kromer earlier this year and will do more next year. Tristan is excellent.

Once Ajabbi gets financially strong, I intend to sign up for weekly paid mentoring and advice from Tristan. It's not cheap, but it will be the best and worth it.

I recently got early access to the Krobar.ai beta from Kromatic. Below are my notes, which I'm adding to as I play.

Resources

References

  • Reference

Repository

  • Home > Ajabbi Research > Library > Subscriptions > Kromatic
  • Home > Handbook > 

Last Updated

05/12/2025

Playing with Krobar.ai

By: Mike Peters
On a Sandy Beach: 05/12/2025

Mike is the inventor and architect of Pipi and the founder of Ajabbi.

Krobar.ai is a simulation platform from Kromatic. It's currently in beta.

Signing up

I signed up for a free trial in early September to test and provide feedback. I then got distracted and forgot about the trial. When I remembered, Meagan Wilder kindly let me back in early November, and I started playing with Krobar.

Experiments

There is no help documentation yet. My experiments started with me pressing buttons and seeing what happened, OK, playing. I then learned how to write ChatGPT 4.1 prompts and slowly built some skills. The chat dialogue appears highly accurate. I usually ask it to use industry-standard data for the default modelling values. I didn't have enough industry-standard values for mean, standard deviation, min, and max to tune the models yet. 😉

One trick I found was to go to Google AI, ask a question, then turn the answer into a Krobar Prompt. Then I might find some of Steve Blank's writing, feed it into Gemini 3, and use the output as a prompt. 😀

Simulation

Each simulation has these pages.

  • Journey Map
  • Workflow diagram
  • Simulation
    • Tornado Chart
    • Histogram
    • Sensitivity Analysis
    • Financial Model (Spreadsheet)

I'm still learning by trying the menu items. There are some hidden tricks. I learned some more by watching the video interview below. The order in which things are done matters with the AI.

Downloads

Everything can be downloaded in various formats. Very useful.

  • PDF Document
  • JPG image
  • Excel Formula
  • CSV data
  • JASON document

Models

These large, complex simulation models were built. Sometimes it took several attempts. They got better as I learned.

  • Ajabbi startup
  • Airport
  • Data Centre
  • Forestry Logging
  • Hospital
  • Movie Studio
  • Shipping (maritime)

I exported the Excel formulas for import into Pipi's industry financial models.

I will do some more soon, with a better understanding of how this works and predesigned prompts, and a bit more cunning. 😊 Then try ARR, ROI, etc.

Suggestions

Direct visual editing of the journey map

  • Enable drag-and-drop editing of journey map connections.
  • Enable more journey map steps to be created and added using a form.
  • Acts as input to AI

Interview with Tristan Kromer about Krobar.ai

Sample outputs from Film Studio 2 Simulation

By ,
Tags , , , ,

No comments:

Post a Comment

Subscribe to: Post Comments (Atom)