What is a Bonsai Sample?

A Bonsai Sample is a demonstrator that can be built to serve one or many purposes. Some examples include:

• Showcasing Bonsai functionality
• Proving Bonsai Brains can outperform prevailing industry methods
• Demonstrating a specific Bonsai use case or vertical
• Showcasing a specific simulator connection
• Demonstrating a Partner’s capabilities

Principles

When building a sample, keep the following principles in mind to make the most impact:

• Reproducible - repeatable results give credibility to the sample and story
• Explainable - keep the sample as simple as possible to make the story as concise and impactful as possible
• Realistic Complexity - make sure to have enough realistic complexity to show a brain outperforming the prevailing industry method. For example, the Chemical Reactor sample simulation includes noise to show when a brain succeeds over a gain scheduled PI controller.
• Maintainable - ensure the sample can be used at scale

Recommended Bill of Materials

A great sample will include all of the following components:

1. AI Spec

This is the brain design document that describes what brain should be built and why. This is often written by a Solution Architect and used by an Applied AI engineer as a starting point to write the brain specification, called Inkling.

• See the AI Spec Template

• Take the free Coursera courses Machine Teaching for Autonomous AI and Designing Autonomous AI to learn how to identify Bonsai use-cases and write an AI spec

• Talk to your Bonsai sales or partner representative to learn about enrolling in the instructor led Cultivate course

• Example: Chemical Reactor AI Spec

• Example: Moab AI Spec

2. Simulation

This is the training environment for the Bonsai brain.

• See simulation documentation

• See Python Samples on github

• See MATLAB Simulink Samples on github

• See AnyLogic Samples on github

• See VP Link Samples on github

• See ROS Samples on github

3. Inkling

This is the Brain specification that defines what and how a brain is trained.

• See Inkling Documentation

• You can learn how to train a brain by taking the free Coursera course Building Autonomous AI

• Talk to your Bonsai sales or partner representative to learn about enrolling in the Bonsai Advance Course

4. Visualizer

The Bonsai App visualizer allows you to create custom visualizations of your simulation that can be viewed in the Bonsai Web UI during brain training and assessment.

• Learn about visualizers and find examples on github

• Example: Chemical Reactor visualizer

5. Benchmark

This is the prevailing industry method that the brain will be compared against to define success.

• Example: Chemical Reactor Sample

6. Test protocol

Documented instructions of how results were obtained so they can be replicated.

7. Results

The data that shows how a brain performs compared to a benchmark

• Example: see the final slide of the Chemical Reactor Story Presentation

8. Sample Story

Storytelling is an important technique for communicating the value of the Sample that you design

• You can learn how to tell a use case story by taking the free Coursera course or the instructor led Cultivate course

• Example: Chemical Reactor Story Presentation

9. Demo Script

A demo script allows others to tell the sample story and demo the sample for others

• Example: Moab Demo script

10. Demo Video

A demo video is a recording of the sample story and demo

• Example: Chemical Reactor demo video

11. Publication (white paper)

Publication of sample results and why they are valuable

12. Azure Marketplace Listing

List the sample in the Azure marketplace so users can discover it

• Learn how to publish and provision an asset in the marketplace here
• Example: CSTR marketplace listing