Atomic KB vs localBots

Atomic is a client for your self-hosted Atomic knowledge base. It connects to your own server to give you mobile access to your notes, saved articles, and semantic search. Save any web page from Safari or any browser using the Share Extension — Atomic automatically extracts the article content, generates embeddings, and connects it to your existing knowledge. Features: - Browse, create, and edit markdown notes - Save web articles from any app via the Share Extension - Semantic and hybrid search across your knowledge base - Hierarchical tag browsing - Offline support with local caching - Connect via QR code or manual setup Requires a self-hosted Atomic server. Learn more at https://github.com/kenforthewin/atomic

localBots gives you a team of AI bots that live on your iPhone. Each bot has a name, a job and a face, and its own thread where you talk to it like a coworker. Start with the built-in team (Sam for general help, Dev for code, Mira for explanations, Ada for writing…) or hire new teammates from templates: a researcher, a planner, a coach, a translator, a chef. Rewrite how any of them works in plain language. GIVE THEM ROUTINES • Reminders that arrive as notifications even when the app is closed • Jobs the bot does for you, once, daily or weekly, with the result posted in its thread • Optional web search before a job, with sources cited 100% ON-DEVICE • Models run on your iPhone's chip with Apple MLX. Your chats never leave the phone. • No account, no subscription, no analytics • Works offline once a model is downloaded CHAT THAT FEELS ALIVE • Streaming answers with Markdown and code blocks • Think mode shows the model's reasoning as "Thought for 4s" • DeepSearch searches the web and cites [1], [2] • Vision: ask about a photo with a vision model • Dictation and hands-free voice chat with any bot MODELS Download open models from Hugging Face in one tap (Qwen3, Llama 3.2, Gemma, DeepSeek R1, SmolVLM) or add any MLX repository. The 0.6B model is 350 MB and runs on every supported iPhone; 3–4B models need 6 GB of RAM or more. Because iOS does not allow apps to run a language model in the background, "job" routines run as soon as you open the app after their scheduled time. Reminders are always on time.