About Us

We are 100% open sourced

Fournex is an AI startup focused on developing techniques for developing Autonomous Agents. We utilize DSPy, a framework for algorithmically optimizing language model prompts and weights. DSPy allows us to treat language models as optimizable components within a larger machine learning system.

By separating the system logic from the model parameters, We enable the use of novel "optimizers" - language model-driven algorithms that can tune prompts and weights to maximize a given metric. Rather than manually prompting and finetuning models through trial and error, we automate this process. Our optimizers can routinely teach models new skills and avoid failure modes through data-driven parameter optimization. This represents a paradigm shift where language models fade into the background as optimized pieces of a system that can continuously learn from experience.

At Fournex, our primary objective is to pioneer the development of powerful and productive AI Autonomous agents. We firmly believe that the ongoing research and advancement as well as frameworks such as DSPy, LangGraph (LangChain) and LLamaIndex model optimization are pivotal in creating robust and secure language model applications. Our focus is on devising techniques that enhance the systematic reliability of models, thereby minimizing the necessity for extensive human oversight.