OpenOrigin: Development of an AI model aggregator

Our experience in MVP development for startup OpenOrigin, an open source platform where developers can host their own AI models, test others' solutions and share experiences.

We needed to create an open platform for developers and AI evangelists with the following features:

• Quick and easy registration of a model in the catalogue;
• Ability to test proposed solutions;
• Collect feedback on products.

For more insights on how to define the functionality of the future product and how to conduct research, read our blog.

The project started with MVP development. We created a database of AI models where developers could post their solutions and users could find handy tools. This approach meets the needs of both sides: authors get a free channel for promotion, and users get access to modern solutions and trends in AI.

We designed an open registration system for the platform so that each author can quickly create an account for their model and add key information: the type of solution and the tasks it performs, a description of the general principle of the product, links to test the technology, and an optional link to test the technology or to the GitHub repository.

We also support the open source philosophy. And we're waiting for you at the link in our GitHub.

Models on the platform can be searched by filter, name, or functionality. Users simply select a use case - the task they want AI to solve – from the suggested list. For example, to find services that automate data processing or image generation. A selection of desired solutions is automatically generated. There are separate filters for models using with Natural Language Processing (NLP) and tables.

After testing the model, users can "Like" their favorite solutions. This makes it easier for authors to gather feedback and helps keep track of popular technologies and products. In the future, comments and commits may be added in the future to support open source principles.