agents

Agents making my life easier.

Core architecture components

1. Foundation Model: Llama 3 as the base LLM
   • Will power all agent reasoning capabilities
   • Can be run locally or via API depending on your compute resources
2. LlamaIndex Framework
   • Provides agent structuring, RAG capabilities, and tool integration
   • Handles context management and query routing
3. Specialized Agents
   • Planning Agent: Coordinates workflows and breaks down tasks
   • Calendar Agent: Manages scheduling and time-based activities
   • Email Agent: Handles email drafting, summarization, and organization
   • Research Agent: Gathers information from various sources
   • Memory Agent: Retrieves past interactions and maintains context
   • Task Agent: Executes specific actions and reports results
4. Shared Memory & Orchestration
   • Maintains state between agent interactions
   • Routes messages between specialized agents
   • Tracks task completion status
5. Chroma Vector Database
   • Stores embeddings for documents, past interactions, and knowledge
   • Enables semantic search for relevant context
6. External API Connections
   • Calendar services
   • Email providers
   • Web search capabilities

Ollama setup

To run locally with Ollama:

1. Install Ollama from https://ollama.com/download or via curl -fsSL https://ollama.com/install.sh | sh (macOS/Linux only)
2. Pull the llama3 (or other) model ollama pull llama3:8b
3. Run the Ollama server: ollama serve