Database teams will manage thousands of AI agents, not individual databases

The role of database administrators is undergoing a fundamental transformation. Rather than spending their days maintaining individual databases, DBAs of the coming years will focus on governing the autonomous agents responsible for database operations at scale.

The AI takeover of database operations#

For decades, database administrators have handled the operational minutiae of keeping systems running: provisioning capacity, diagnosing slow queries, managing migrations, and responding to outages. AI agents are now assuming these responsibilities, moving beyond simple automation scripts to make decisions based on real-time conditions.

When an agent encounters a performance issue, it can investigate metrics, identify the problematic query, decide on a response, and verify the outcome—all before a human intervenes. This doesn't eliminate the DBA role; instead it elevates it. Database teams transition from performing individual tasks to defining what agents can do autonomously, where human review is necessary, and how to handle errors.

Scale that demands new infrastructure#

The numbers underscore why this shift matters. Gartner forecasts that Fortune 500 companies will operate more than 150,000 AI agents by 2028—a leap from fewer than 15 today. Yet only 13% of organizations feel they have adequate governance frameworks in place.

Each agent will likely need its own data store for state management, retrieval, and inter-agent communication. Agent workloads also behave differently from traditional applications: they provision rapidly, remain idle for extended periods, and spike unpredictably when activated. Hiring 150,000 database administrators to manage them individually isn't feasible.

Consolidation as both operational and economic necessity#

Instead of provisioning databases one-by-one, vendors like Yugabyte are promoting fleet management approaches. Their Agentic Multitenant PostgreSQL (AMP) packs hundreds of small workloads onto shared infrastructure while maintaining isolation, letting agents orchestrate lifecycle operations like provisioning and scaling through standard APIs.

This approach addresses cost as much as operations. Agent workloads are notoriously bursty and idle, making traditional capacity planning wasteful. Serverless models with scale-to-zero capability let companies pay only for what they use—agents sitting idle consume no resources. Resource governance prevents any single agent from monopolizing shared infrastructure.