SmallMinds

SmallMinds is a little company dedicated to creating human-editable and human-readable Machine Learning (what some call "AI") models. We have the nerve to believe that the Boldly Bloated "AI" models of today will become a thing of the past in the next 3-5 years, and with courage and vigor are pursuing what comes next. Our first product suite, DataFlood, models your data in seconds and can generate thousands of synthetic test data documents per second. The DataFlood software suite is available to deploy anywhere you need it -- on premise, in the cloud, or both. It never requires database connectivity or any other outside resources to operate and thus is ideal for use in walled-garden environments. Humans get to determine the future of "AI". LLMs and all other ML models are just software, and as such we can determine how these will be developed, deployed, and sustained. SmallMinds believes in human-scale ML that can be edited by humans and thus controlled by humans to solve human problems. We believe that when it comes to ML, less is more. We measure the success of our models on a unique set of metrics: 1.) How small is the ML model? Smaller is better. 2.) How much data is required to train the ML model? Less is better. 3.) How much energy is required to run the ML model, for training and inference? Our goal is 90w of energy per day or less. 4.) Can the model be "taught" by a human, through directly editing the model itself, as well as be "trained"? The answer should always be yes. We are actively seeking to find better ways to do ML that combine deterministic and non-deterministic methods in the same model. There are times when we need "random", when we want the model to do the work. But there are also times when we want full control over what a model will output. And everything in-between. DataFlood meets all of these standards. More is coming. Stay tuned!