Jozu
Self-hosted and fully air-gap capable. Jozu verifies every model, agent and MCP server before it runs, enforces policy on your own Kubernetes, and proves it with one signed audit log.
jozu.comSOFTWARE AND DATA ENGINEERING
Let a coding agent work on your project without reaching the AWS credentials on your Mac. Isolate it in an Agent Guard microVM and pass in only what it needs.
MAC OSCODEX
ARTIFICIAL INTELLIGENCE
Jozu Agent Guard runs each AI agent in its own microVM on Kubernetes, admits only signed ModelKits, enforces tool policy, and keeps a signed audit log.
AGENT ENCLAVEJOZU AGENT ENCLAVE
ARTIFICIAL INTELLIGENCE
Every AI-ready data framework says what ready should mean. None records that anyone checked. Here's how to turn a dataset's label into a signed claim on its digest.
OAK RIDGE NATIONAL LABORATORYCEOS
ARTIFICIAL INTELLIGENCE
For the past two years, the conversation around prompt injection has centred on the model. Better alignment, stronger system prompts, safer fine tuning, and output guardrails have all aimed to make language models more resistant to malicious instructions. Those improvements have made a real difference, yet prompt injection remains one of the most persistent security
ARTIFICIAL INTELLIGENCE
Learn how speculative decoding pairs a small draft model with a large target model to speed up local LLM inference, and how to run it with Jozu Rapid Inference Containers.
JOZU