Redefining the Interface for the Future of Computing
Decades ago,
personal computing was a mere vision in imaginative minds. The founders of Microsoft, Bill Gates, and Paul Allen, visualized how microprocessors and language transforms would shape the future of computing. They were early pioneers, recognizing the massive potential of a relatively unused piece of technology – the microprocessor.
At BRX AI,
we see a similar shift in technology represented by generative AI, specifically large language models (LLM) and situation learning models (SLM). We view this as the evolution of the microprocessor revolution, a new sea change in technology that will redefine the future of computing.
Three years ago,
when GPT-3 models were first released, it felt like a monumental leap forward. BRX AI was there at the outset, fascinated by the enormous potential of these models. We started experimenting, interacting with LLMs, witnessing the profound paradigm shift in how humans can seamlessly interact with computers.
Similar to how Microsoft built on the idea of the microprocessor to create the PCs we know today, we at BRX AI are building the next generation of computing systems. We are constructing a computer, an interpreter, and a new coding language for AI. We call it the BRX engine.
Our journey took us through the struggles of early-stage adaptation and implementation. The pandemic-induced shift to online education gave us testing grounds for our first LLM applications, Cheat.school and Studywith.ai now merged to BRX.ai. Our pioneering work in prompt chaining helped us lay the groundwork for where we are today.
Learning from these experiences,
we are aiming to design and construct the infrastructure that fills the 'inference gap' in generative AI. Our primary thesis at BRX AI is that "every single token counts." In a world where models act as hyper-deterministic machines, their inference is as important as their output for achieving high-quality results.
Most of the current AI landscape focuses on developing agents. We've observed that this approach leads to inevitable problems – it presumes AI can function entirely independently. It negates the symbiotic link that exists between humans and AI, where humans bring much-needed entropy to hone AI performance.
At BRX AI,
we believe in a world where humans and AI safely and securely compliment each other. We imagine a future where we leverage this harmonious relationship to foster creativity and intelligence.
We are proud to introduce the BRX engine V1 and the upcoming V2. They form the foundation of our pursuit towards better interaction between humans and AI. The BRX engine V1 has already benefited from over 100k uses, proving its stability. It has exhibited capabilities that extend beyond the limitations of the standard tail-head agent approach.
The BRX engine transitions
us from a deterministic machine learning approach to one that acknowledges and appreciates the discrete nature of these models. V2 is set to expand this capability even further. With an added ability for the LLM to parse a high-level interpolated language within the engineering, the BRX engine V2 will offer unparalleled flexibility and adaptability.
We foresee these engines playing an integral part
in the new era of AI development. They will empower developers to do so much more than they currently can with generative AI while also heralding a new era in rapid prompt engineering.
In essence, we're attempting to secure the future by improving upon the present. We're building the tools that enable developers to reinvent how we interact with technology, akin to what Microsoft achieved with microprocessors.
We're inviting you to be part of this exciting journey
into the future of computing. As we spark this revolution, let's remember that computing has always been about enhancing creativity, not replacing it.
Welcome to BRX AI. Welcome to the future.
