ARGO
A local AI agent for Windows, macOS, Linux and Docker, with Ollama models, offline document search, research tools and connections to cloud model APIs.
ARGO is a local AI agent platform for people who want assistants that can research a topic, work with their documents and carry out tasks with several steps. It runs on Windows, macOS and Linux, including Intel and Apple Silicon Macs, or through Docker. You can create specialized assistants by describing their role in plain language.
Its research engine uses multiple agents to break a request into steps, call tools and produce reports. You can revise a research plan in natural language before execution. Built-in tools cover web search, scraping, browser control and local file management; MCP support lets assistants use additional tools.
The local knowledge base accepts files, folders and websites. It reads PDFs, Office documents, Markdown and plain text, keeps linked folders synchronized, and traces answers to reference passages. For complex questions, it breaks the query into smaller parts and checks whether it has enough information to answer.
ARGO works with Ollama and GGUF models from Hugging Face. It also connects to OpenAI, Claude, DeepSeek and services with an OpenAI-compatible API, with model switching available during a conversation. Local operation supports offline use and keeps stored data on your machine. Choosing a cloud model sends requests to that provider, while web research needs internet access.
The stated minimum hardware is a four-core CPU and 8 GB of RAM. Docker deployments can use NVIDIA CUDA with the NVIDIA container toolkit.
