Oumi automates custom AI model development for enterprises
Oumi PBC introduced a platform today designed to automate the creation of custom AI models, targeting enterprises moving away from large, general-purpose systems toward specialized alternatives.
The startup, which operates as a public benefit corporation, built the platform on an open-source project that has accumulated nearly 9,000 positive ratings on GitHub and adoption across dozens of research institutions. The new offering extends that work to enterprise teams that lack the time or expertise to build models from scratch.
The shift away from large models
Enterprise teams increasingly want to replace off-the-shelf large language models with small language models (SLMs) tailored to specific tasks, said Manos Koukoumidis, Oumi's chief executive. These specialized models offer greater relevance to particular projects, lower costs, and faster response times.
The transition has stalled because building custom models traditionally takes months and requires deep technical knowledge. Oumi's platform addresses that by automating the entire workflow-data generation, evaluation, training, and iteration-that engineers would otherwise handle manually.
Automation claims
Oumi says its system can build custom models up to 100 times faster than conventional processes, reducing weeks or months of work to hours or minutes.
Users define a task in natural language rather than writing code. The platform then automatically defines evaluation metrics, generates synthetic data, fine-tunes models, and iterates based on performance gaps. "It analyzes the results for you, automatically synthesizes data, fine-tunes the model and keeps iterating," Koukoumidis said.
The system supports a range of open models. Pricing is based on compute usage and tokens for training and inference, though Koukoumidis said the cost is offset by reduced engineering effort.
Broader strategy
Oumi frames the platform as part of an effort to decentralize AI development. The company's goal is to "democratize the use of AI and put the future of AI in the hands of enterprises," Koukoumidis said.
For development teams evaluating custom model approaches, understanding how generative AI and LLM development has evolved can provide context for these options. Teams interested in practical implementation should also explore AI for IT & Development resources.
Your membership also unlocks:
