Sevra
Private AI app that runs locally on your computer with automatic model
Visit sevrahq.com ↗What is Sevra?
Sevra is a local AI application designed to work with your files and tools while keeping data private on your device. Built on Slotstream, an open-source engine that runs a 105 GB model on Macs with 16–64 GB of memory by loading parts from the SSD as needed. The app uses native Swift on Apple's MLX and Metal, optimized as a single system for the model.
Features include searchable conversation chapters with a personal journal, file and tool integration, and the ability to create custom tools or save repeatable tasks as skills. Sevra automatically selects and updates models based on your hardware, testing answer quality, speed, and tool use on similar devices. Currently in development for Mac first, with Windows and Linux planned. Works offline after the initial model download.
Momentum
Daily activity over the last 30 days: upvotes, unique visits, new GitHub stars and Hacker News points.
How do we know it's vibe coded?
vibetrend discovered Sevra automatically. Evidence: Agent files (claude-code) in carloslfu/slotstream. View source.
Similar vibe-coded apps
#2 this week
Manage teams of AI agents to run your business from one dashboard
Agent development environment for running multiple coding agents in parallel
AI converts documents into natively editable PowerPoint presentations
Open-source video production system directed by AI agents
Open-source AI job search agent that runs locally in your CLI
Open-source desktop client for DeepSeek Harness with plugin architecture
Open-source Android agent that runs on-device with no root required