- Autonomous agents that can reason, plan, use tools & execute
- Multi-agent collaboration (swarms, crews, hierarchies)
- Long-term memory + episodic memory for LLMs
- RAG that actually works in production
- Local 70B+ models beating cloud APIs
- Turning any idea into running code with zero human in the loop
Core β’ LangChain β’ LangGraph β’ CrewAI β’ AutoGen β’ LlamaIndex
LLMs β’ GPT-4o β’ Claude 3.5 β’ Llama-3.1-70B β’ Grok β’ Mixtral
Local Runners β’ Ollama β’ vLLM β’ LM Studio β’ Oobabooga
Vector DBs β’ Chroma β’ Pinecone β’ Weaviate β’ Qdrant
Tools β’ Tavily β’ Browserless β’ Playwright β’ Function calling
Languages β’ Python (main) β’ pytorch tensorflow