AI agents are moving beyond simple chat. Building them well means understanding how they reason, use tools, remember context, and recover when work gets complicated.
Join MongoDB for a concise, three-part workshop series on the architecture behind agents that can do more than respond. Learn from practitioners and bring your questions to the live sessions.
October 15 — Build the foundations of AI agents: Move beyond simple chat and enter the year of the agent.
- Learn the reasoning, tools, and memory architectures required to build autonomous, learning agents.
- Explore agent architectures and production-grade workflows.
- Build a controllable MongoDB learning agent.
October 22 — Design memory systems: An agent is only as good as its memory.
- Learn a practical framework for architecting memory systems that maintain context, learn from experience, and handle long-running tasks.
- Identify storage patterns for persisting and structuring memories.
- Implement retrieval strategies using metadata and recency signals.
- Manage the memory lifecycle to maintain system performance.
October 29 — Inside Atlas Agent Engine: Memory, Human-in-the-Loop, and MCP Tools you already use.
- Overview and demos: A platform overview, followed by two reference agents for insurance and travel.
- Core capabilities: Long-term memory across sessions, human approval for risky actions, and tools connected over MCP.
- Key distinction: Execution state (checkpoints for resuming failed or paused runs) is kept separate from long-term memory (what an agent deliberately retains for future sessions).