London-based AI startup Mindstone has launched Rebel, a local-first agentic operating system that automatically selects the best AI model for any given task. The platform addresses one of enterprise AI’s persistent headaches: the complexity of choosing and switching between models.
The system uses a markdown-based shared memory architecture to route workflows between local and cloud-based models, optimizing for both data security and cost. Teams of up to 100 users can deploy the platform under a “Fair Source” license, which Mindstone positions as a middle ground between open source and proprietary software.
“Enterprise AI agents should automatically remember which model is right for which task,” said Mindstone’s team in their announcement. The platform learns from interaction patterns and builds a model selection memory over time.
Rebel is available for macOS and Windows, with Linux support in development. The local-first approach appeals to enterprises with strict data residency requirements, as sensitive workflows can stay on-premises while still benefiting from model orchestration.
The launch comes amid a crowded landscape of enterprise AI agent platforms, but Mindstone differentiates through its emphasis on local deployment and automatic model routing — reducing the manual configuration typically required for multi-model AI stacks.