# What Are Cloud Experiments

Please continue reading if you are interested in contributing to this repository, building your own Cloud Experiments, or just curious how Cloud Experiments are distinguishable from samples, starter apps, and tutorials. As source we could build a Cloud Experiment from ground up, reuse an existing sample code, or build on top of a starter toolset. As destination we could target a blog-like tutorial for building the Cloud Experiment on AWS Console, create a Streamlit exploratory data app, author an Amazon SageMaker notebook, design a single page interactive data story, or build a starter app using AWS Amplify along with reusable components in Flutter or React. Constituents of a Cloud Experiment usually include a working starter app, identified public dataset(s) suitable for the experiment, a set of reusable custom components for a specific set of use cases, and an API which relates to problem domain or solution design workflow. Components and API play an important role in building explainable solutions which in turn encourages collaboration, open innovation, and adoption. Cloud Experiments can be used as do-it-yourself, however they are designed for collaborative exploration between developers, designers, and domain experts. Cloud Experiments get better with use as collaborative exploration generates new feature ideas, discovers new datasets, or expands the set of use cases. Popular Cloud Experiments may be made production reusable, available in AWS marketplace for reuse by real-world applications. This in turn creates a feedback loop for improving the base Cloud Experiment.

Following variants of Cloud Experiments increase in investment of time to create, multi-skills required, and value they mght deliver. Many Cloud Experiments are a mix of these variants or evolve from one variant to the next over time and reuse.

**Art Of The Possible:** Cloud Experiments could also focus on demonstrating the art of the possible with AWS technologies. A collection of demos for AWS service(s) accompanied by a reusable set of components or API abstractions.

**Starter App:** Cloud Experiment can work like a starter app for a specific use case, bringing together the optimal combination of code dependencies, developer packaging, developer workflow automation, building block components, and sample code tailored for a specific use case.

**Exploratory Workflow:** Cloud Experiments are about exploring the problem-solution space between developers, designers, and domain experts. We want to code as close to describing the exploratory workflow for problem discovery and solution definition. This means extracting API which describes activities in the workflow. The API should ideally deliver no code interactive experience in the app to facilitate exploration. As an example, if there is code to source data from a remote repository, turn this into an API which renders UI component to expect user input and displays the data for exploration. This makes the experiment reusable across multiple datasets and use cases. This also makes writing or iterating the Cloud Experiment app a low code experience.

**Interactive Data Story:** Really good Cloud Experiments tell an interative story with data. Think data journalism like Guardian or New York Times newspapers. These are well researched stories, into a particular problem-solution domain, backed with real-world open dataset(s), explorable with interactive visualisations, and complete with an innovation story moving from problem definition to scaling idea to solution demonstration.

**Codify Domain:** Ultimately Cloud Experiments codify domains related to design methods, innovation techniques, and industry body of knowledge.