Readyset Launches QueryPilot Guard, a Control Plane for AI-Generated Database Queries

SAN FRANCISCO — Readyset, the database scaling company building for the AI era, today announced QueryPilot Guard, a new control plane that lets enterprises safely run AI-generated SQL against production databases. QueryPilot Guard sits between AI agents, copilots, and analytics tools and the databases they query, giving platform teams the visibility, policy enforcement, and cost guardrails they need as machines—not humans—become the dominant author of database traffic.

AI is rapidly changing how organizations interact with data. Queries are no longer carefully reviewed and tuned by engineers; they are generated on the fly by language models, embedded copilots, and autonomous agents. The result is a workload that is more dynamic, more exploratory, and dramatically more frequent—one that hits production systems in real time without human review. Most modern databases were never designed for this. Scaling them is no longer just a performance problem. It is a control problem.

QueryPilot Guard gives platform, data, and security teams a single layer to manage AI database access at scale. Capabilities include policy-based query admission, automatic detection and rewriting of expensive or unsafe patterns, per-agent rate limiting and budgets, and redaction of sensitive columns. It deploys in front of Postgres and MySQL without application code changes and builds on Readyset's caching and incremental view maintenance technology to keep latency low even as AI traffic spikes.

"The assumption that a person wrote the query is quietly disappearing from modern data stacks, and the controls built on top of that assumption are breaking with it," said Gautam Gopinadhan, CEO of Readyset. "QueryPilot Guard gives enterprise teams a way to say yes to AI on their data without giving up reliability, safety, or their budget. It's the missing layer between every AI tool in your stack and the database underneath it."

Early customers are using QueryPilot Guard to cap runaway agent behavior, prevent regressions from ad-hoc model-generated SQL, and produce audit trails that satisfy internal governance and external compliance reviews—without renegotiating access for each new AI tool added to the stack.

Availability. QueryPilot Guard is available today in private preview for enterprise customers, with general availability planned for later this year. Teams interested in early access can request a briefing at https://readyset.io/.

About Readyset

Readyset is a database scaling platform built on a decade of research at MIT and developed by a team of SQL query experts with deep knowledge of how queries behave in production. That expertise has helped global engineering teams scale their Postgres and MySQL infrastructure through automated caching and performance optimization - without application code changes. Now with QueryPilot Guard, Readyset is applying that same foundation to a new mission: making databases ready to scale safely and securely for the AI era. For more information, visit https://readyset.io/.

Media Contact

info@readyset.io