AI / LangGraph LangChain Interview questions
How do you integrate memory with agents?
By default, AgentExecutor has no memory — each invocation is stateless. To give an agent conversation memory, pass a memory object to AgentExecutor. This is distinct from return_intermediate_steps (which stores tool call history within a single run); memory stores the dialogue across multiple separate invocations.
from langchain.memory import ConversationBufferMemory from langchain.agents import AgentExecutor memory = ConversationBufferMemory( memory_key="chat_history", # must match prompt variable return_messages=True, ) executor = AgentExecutor( agent=agent, tools=tools, memory=memory, verbose=True, ) executor.invoke({"input": "My name is Alice."}) executor.invoke({"input": "What did I just tell you?"}) # recalls "Alice"
The prompt used by the agent must include a {chat_history} variable (or whatever memory_key is set to) so the history is injected on each call. For multi-user scenarios, each user needs their own memory object — or use LangGraph's checkpointing with thread IDs to manage per-conversation state.
What must the agent's prompt include for ConversationBufferMemory to work correctly?
How is conversation memory different from return_intermediate_steps in AgentExecutor?
More Related questions...
