The Internet is increasingly becoming a collection of closed platforms. A reliable exchange between them is a tough and complex endeavor for independent software developers and small teams.

Homunkulo is committed to find solutions for this increasing problem. Can machine learning help construct, maintain and sustain connections between closed platforms? Is there a format as ubiquitous as the spoken or written language, in which services and users can exchange data?

The architecture is based on three cornerstones:

• A format with which the problem domain can be described
• A software that uses Machine Learning to compare this definition with the formats to be connected
• Automatically generated software libraries which allow the user to connect services without knowing about their specific implementation details

Homunkulo is Open Source and licensed under the MIT License. It uses Python for the Machine Learning aspects and Go for the automated generation of code. Ideally, it will be flexible enough to serve any target language.

Feel free to contact us with ideas, questions or requests.