Sanctum is a cutting-edge AI assistant application designed to run full-featured open-source Large Language Models (LLMs) locally on your Mac device. It prioritizes user privacy by using AES-256 encryption to secure data, including chat history and caches, ensuring no data leaves your device. Sanctum allows users to import various document formats like PDFs and DOCX, enabling them to ask questions, get summaries, and interact with AI in a completely private manner. It is suited for anyone needing a secure, reliable AI assistant for personal or professional use.

GAMA Genstar Plugin adds generative AI capabilities to the GAMA platform by providing connectors to OpenAI, local LLMs, and custom model endpoints. Users define prompts and pipelines in GAML to generate agent decisions, environment descriptions, or scenario parameters on the fly. The plugin supports synchronous and asynchronous API calls, caching of responses, and parameter tuning. It simplifies the integration of natural language models into large-scale simulations, reducing manual scripting and fostering richer, adaptive agent behaviors.

Ollama Bot is a Node.js-based AI agent designed to run on Discord servers, leveraging the Ollama CLI and local LLM models for generating conversational responses. It establishes a persistent chat context, allowing users to maintain topic continuity over multiple messages. Administrators can define custom prompts, set model parameters, and restrict commands to specific roles. The bot supports multiple LLM models, automatically manages message queues for high throughput, and logs interactions for audit purposes. Installation involves cloning the repository, installing dependencies via npm, and configuring environment variables such as the Discord bot token and Ollama settings. Once deployed, the bot listens for slash commands, forwards queries to the Ollama model, and posts generated replies directly in Discord channels.

echoOLlama leverages the Ollama ecosystem to provide a minimal agent framework: it reads user input from the terminal, sends it to a configured local LLM, and streams back responses in real time. Users can script sequences of interactions, chain prompts, and experiment with prompt engineering without modifying underlying model code. This makes echoOLlama ideal for testing conversational patterns, building simple command-driven tools, and handling iterative agent tasks while preserving data privacy.