Ollaya – Ollama for open-source, Jev-style decision models

You can now run AI decision models locally without per-token fees, keeping your data private on your own hardware.

Ollaya has launched a new open-source platform that allows users to run decision models locally on their own hardware, offering a private and cost-effective alternative to cloud-based services. The platform is designed for fast, typed decisions, returning calibrated answers in milliseconds. It supports models such as Laya from Convai Innovations, Decider, NLI, and Gliclass, with more planned. Ollaya aims to be a drop-in replacement for TypeSafe's API, enabling developers to use the official TypeSafe Python SDK or call compatible endpoints directly by setting environment variables. This local execution means user data remains on their machine, addressing privacy concerns. Ollaya is available as a desktop app and command-line interface for macOS, Windows, and Linux, with GPU acceleration supported for significantly faster processing.

Ollaya’s introduction of local decision model execution directly challenges the established cloud-based API model for tasks requiring structured, fast answers. By enabling users to run these models on their own GPUs, it bypasses the transactional costs associated with cloud APIs, positioning itself as a cost-effective and private solution for handling sensitive data. The platform's compatibility with TypeSafe's API lowers the barrier to adoption for existing users and developers in that ecosystem, suggesting a potential shift towards decentralized AI inference for specific, predictable workloads.

The emphasis on speed, measured in milliseconds for local execution compared to potentially higher latency for hosted APIs including network time, highlights a key differentiator for time-sensitive applications. Furthermore, the provision of calibrated probabilities allows for more reliable decision-making by enabling the setting of confidence thresholds. This focus on both performance and reliability, coupled with the open-source nature and local data handling, could influence how businesses evaluate and deploy AI solutions for internal processes and customer interactions.