“I want to build a multi-agent system that researches a topic, writes a summary, and then has another agent fact-check it. Walk me through the architecture, the system prompts for each agent, and the orchestration code in Python.”
“How does LangGraph’s StateGraph differ from hand-rolling my own agent orchestration in plain Python? What specific problems does it solve, and when is the extra complexity worth it?”
“My multi-agent pipeline is producing inconsistent output because agents are passing information in incompatible formats. What patterns can I use to standardise handoffs between agents — both in the system prompts and in the Python orchestration code?”
“I want to run 8 research agents in parallel using the OpenAI API without hitting rate limits. Show me how to implement concurrency with a semaphore to cap simultaneous requests, and how to handle partial failures gracefully.”
“What is the right number of agents for a task I’m designing? Give me a decision framework for deciding between one agent with more tools, two agents with a handoff, and a full multi-agent pipeline.”
