Ben Cefalo describes MongoDB's database product decisions as a response to customer feedback across startups and enterprises. He explains how agents change workload intensity through bursts of requests and delegated tasks, even when the underlying operations remain familiar.
Ben Cefalo discusses separating compute and storageAI compute infrastructure is the hardware, facilities, networks, power, cooling, storage, and software used to train and run AI models. in Atlas Infinite so customers can scale resources more independently. He contrasts this with per-query chargingUsage-based pricing charges customers according to measured consumption of a service rather than only a fixed recurring fee. and describes improved price-performance as a way to create workload headroom. The product performance and pricing claims are attributed to his account.
Ben Cefalo says production adoption also depends on security, governance and observabilityGovernance defines decision-making authority, rules and accountability within a system or organization.. He highlights unresolved questions about liability for agent actions and whether audit trailsAn agent audit trail is an ordered, durable record of an AI agent's instructions, decisions, tool calls, state changes, identities, and outcomes. can reliably prove what an agent did. The conversation treats these as operational challenges rather than claiming that smarter models alone solve them.
Watch on YouTube




