AI

For now, we are using the latest LLM model from OpenAI. We use retrieval-augmented generation (RAG) to ensure that the AI is always up to date with the latest information from your lab. This allows the AI to provide accurate and relevant responses based on the most recent data from your lab.

Every time you add a new object to Liminal, the AI is updated with that information. For example, if you add a new sample, the AI will know about it and be able to answer questions about it. Say you add a knowledge note to your agarose powder that "users must secure the lid tightly to prevent drying out."" The AI will remember this and be able to answer questions about it in the future. This can even turn into an alert if another user references Agarose in their own notebook entry!

Data Storage

Right now, all Liminal data is stored on secure cloud storage. We are working on allowing you to choose your own storage provider. Users will be able to "plug in" their own storage provider, such as the cloud storage providers your University already supports or a University compute cluster.

For example, at Purdue University, where the first labs are using Liminal, we are working to integrate with the Purdue Research Computing storage system. This will allow researchers to use their existing storage without worrying about moving data around—making Liminal more affordable and easier to use.

Compute Power

Currently, all Liminal computing is handled on a cloud server. However, we are working on allowing users to "plug in" their own compute source, much like data storage. This will enable users to leverage their own compute resources, such as a University compute cluster or cloud-based compute options.

Login and Authentication

We are working on allowing users to log in using their existing University credentials. This integration will simplify access, eliminating the need for additional passwords, and make it easier for University IT teams to manage permissions.

Alternatively, users can log in with a Google account or create a separate Liminal account for authentication.

Data Provenance & Reproducibility

Liminal consists of various "entities" or "objects" representing different parts of your lab. The platform is designed to interconnect these entities, giving you a complete view of how everything in your lab is related.

The **notebook** serves as the hub for tracking and referencing relevant objects, ensuring data provenance and reproducibility in your research.