Helping people make better decisons

We're building the first AI Agent focused on autonomous data analysis.

Here are some of the principles that guide our development.

Everything starts with understanding data

We started in 2017 with a focus on building an AI that can traverse data and represent complex data operations with simple abstractions.

This "Linguistic Data Layer" is far easier for LLMs to use and allows for more complex and trustworthy agents.

Data and language require different AIs

There will be as many forms of machine intelligence than there are of human intelligence. Just as humans specialize and work together, great AI systems will use a variety of methods intelligently combined.

At Unsupervised, we use LLMs for natural language tasks and for most reasoning tasks, like planning and orchestration. For tasks related to understanding or accessing data, we use an ensemble model of unsupervised learning techniques to understand the data and represent it in a linguistic data layer.

AI Agents and natural language interfaces are the easiest to adopt

Agents communicating in natural language feel far more natural to work with than other approaches. AI Agents using natural language interfaces allow people to work with AI the same way they'd work with remote co-workers.