TuiML builds machine learning libraries designed for autonomous AI agents
TuiML: Machine Learning for AI Agents
Artificial IntelligenceMachine Learning
Summary
Machine learning tools like Weka and scikit-learn were made for human programmers, which makes it hard for AI agents to use them efficiently. The authors created TuiML, a special machine learning library designed so AI agents can understand, explore, and build with it all by themselves. This library includes built-in algorithms and clear descriptions that help agents find and validate tasks automatically, making experiments easier to trace and repeat. Unlike traditional tools, TuiML keeps the data and models private on each device and remains competitive in prediction quality.
What this means in practice
- •For ai developers: Build AI agents that autonomously choose and run machine learning workflows using a library designed for agent understanding and extension.
- •For data science teams: Create reproducible machine learning experiments with automatic validation and tracing of all steps within a single framework.
Authors
Nilesh Verma, Nick Lim, Albert Bifet, Bernhard Pfahringer
Abstract
Machine-learning libraries such as Weka and scikit-learn were designed for human programmers. Language-model agents now use these same libraries by recalling APIs from memory and writing code, an approach that hides what a library offers, delays errors until runtime, and loses experimental state between turns. We present TuiML, a self-contained machine-learning library built for AI agents, with native algorithms across supervised, unsupervised, time-series, data handling, tuning, and evaluation tasks. Every component describes itself through machine-readable metadata and parameter schemas, so an agent can search the library, inspect components, compose validated workflows, and register new ones that become discoverable in turn. Every call is validated, seeded, and traced, and sessions export as runnable notebooks, making experiments reproducible by construction. One specification layer drives the Model Context Protocol (MCP), agent-framework adapters, a Python API, a CLI, and local model serving, while data and models never leave the machine. Benchmarks show TuiML remains predictively competitive with scikit-learn and Weka. While looking like a conventional library to a human user, TuiML is designed for agents first, allowing them to read, extend, and operate machine learning autonomously. TuiML is open source, with documentation at https://tuiml.ai.