Build Your First Chatbot With Memory Using Matrix Context

This article provides a tutorial on building a chatbot with memory using the open-source, local-first Matrix Context engine in Python. It explains how to give a chatbot memory that remembers facts and conversation context across turns, using just three lines of code. The memory is stored as typed context experts (profile, semantic, session, policy) and routes questions to relevant experts before retrieval. The tutorial includes a working example with a fake model that runs without API keys, demonstrating how to add facts, retrieve context, and record turns. It covers installation, basic usage, and the core pattern of two calls per turn: memory.context_for() before answering and memory.record_turn() after.