Ollama: OpenJarvis v1.0 for personal AI agents that run locally

Ollama has released OpenJarvis v1.0, an open-source framework for building personal AI agents that run on your own hardware. The framework was developed by the Hazy Research and Scaling Intelligence labs at Stanford University, and it puts local processing ahead of cloud dependence.

What does the local-first approach mean?

Local-first is an approach in which execution on your own device is the default, and the cloud is optional. As the announcement states, “models run locally and the cloud is optional”. This addresses the questions of privacy, efficiency and cost that are inherent in cloud-dependent systems, since data does not have to leave the user’s device.

What features does v1.0 bring?

Version 1.0 includes native Ollama integration for managing models and performance tracking that measures energy, cost and latency alongside accuracy. It comes with ready-made agent presets — a morning summary, a research assistant and a coding agent. It also supports flexible model selection through the Ollama library, as well as local document indexing and web research.

How do you get started?

Installation is simplified to a single command for macOS and Linux, while Windows is supported through WSL2 or a desktop application. Users can immediately experiment with models or set their own defaults through a configuration file. The framework shows that personal AI applications can run efficiently without a constant connection to the cloud, with transparency about operational costs and environmental impact.