OUR PRODUCT
Introducing CoSMIC
The Cognitive System for Machine Intelligent Computing.
This innovative project aims to create a seamless platform that integrates various AI tools, all powered by Small Language Models. These models are designed to interpret and explain predictions or results from advanced AI tools in plain English.
With CoSMIC, stakeholders can effortlessly analyse complex results provided by experts, making data-driven decisions more accessible and understandable.
ABOUT CoSMIC
How it works
Better together
Integrate different specialised AI predicative tool with large language models (LLMs).
The way forward
Smart routing system integrated within the query analyser.
Chain-Of-Thought
Using Chain-Of-Thought technology to gain the most from LLMs.
Version 1 is ready now
The project has reached its first stable version, successfully demonstrating its capabilities by mastering chess as a proof of concept. The platform can serve as a backend microservice, empowering applications that require reasoning and decision-making processes in plain English. It has numerous applications across various industries, including medical and financial sectors.
We invite you to contact our team to explore and develop additional use cases tailored to your needs.
The future is bright
Our team of researchers and developers is currently working on a new use case: building a code developer assistant. This exciting project is being funded by NetApp Australia. Additionally, we are exploring innovative methods to fine-tune models and employing a more intelligent query analyser.
Our partners
Join us in fuelling innovation!
As a not-for-profit institute, OpenSI is on a mission to revolutionize digital transformation through open-source solutions. We’re seeking financial support to launch the OpenSI grant program, empowering PhD candidates to tackle real-world challenges and drive impactful change. Your contribution can make a tangible difference in shaping the future of technology. Get involved today and be a catalyst for innovation with OpenSI.
