Automatic caching for OLTP, zero code changes.

Today, we're announcing the Public Beta of Readyset QueryPilot, a significant step toward making OLTP workloads, especially AI-generated applications, scale seamlessly. Think of this as the “Look Ma, no hands” moment for database performance. You route traffic once, QueryPilot takes the handlebars, caches safely, and accelerates performance without code changes or manual rules. I've spent a lot of time reflecting on why this matters deeply to me, our team, and the industry, and I'd like to share those thoughts with you.

At Readyset, we're singularly focused on solving the complex challenges of scaling relational, operational workloads, what most folks refer to as Online Transactional Processing (OLTP). These workloads power the critical, real-time applications that underpin nearly every user-facing system today: e-commerce sites, financial services, SaaS platforms, and increasingly, AI-generated applications. Over the past six months, one request has dominated customer conversations: automatically find the bad/high‑impact queries and optimize and cache them without manual effort. That is why we built QueryPilot. It automatically detects high-impact queries in your live workload and accelerates them safely.

OLTP systems must balance low latency, consistency, high concurrency, and relentless availability. Traditional solutions like read replicas, vertical scaling, or manual query optimization have historically solved these problems, albeit with significant operational complexity and high cost. Yet, with the explosion of generative AI tooling, we're seeing a seismic shift: developers are rapidly deploying AI-generated apps built with modern frameworks that automate away most SQL queries, leaving databases inundated by unpredictable, complex query patterns. The volume and variability of queries make manual tuning impractical, reinforcing the need for automatic query caching that QueryPilot delivers.

I predict that within the next few years, nearly all new OLTP-based applications will be at least partially AI-generated. Why? Because AI tooling accelerates shipping dramatically. Teams using frameworks like Rails/ActiveRecord, Django ORM, or Prisma already benefit from this productivity boost today, but the AI wave will amplify it exponentially. More code will be auto-generated, more queries abstracted away, and more databases pushed to their limits.

But there's a critical bottleneck: how do you scale databases instantly, automatically, without manual tuning, even as applications evolve unpredictably? Traditional caching methods like Redis or Memcached, while effective in controlled environments, require manual cache keys, invalidation rules, and significant engineering overhead. They simply can't scale at the speed of AI.

That's precisely why we've built Readyset QueryPilot.

QueryPilot is a fully managed proxy solution that sits transparently in front of your database, automatically analyzing and routing queries to the Readyset cache to cut latency and reduce load. It works particularly well with stable, parameterized query patterns common in today's application frameworks, enabling vibe-coded applications to scale effortlessly. QueryPilot can be deployed on Readyset Private installations as well.

More than a simple caching layer, Readyset encompasses both a high-performance cache engine and a smart proxy. QueryPilot controls the flow, sending eligible queries to the engine automatically and delivering performance gains that once took weeks of manual tuning and dedicated DBA effort.

Why Readyset is built for OLTP

OLTP is the hard problem: low latency, consistency, and concurrency under constant change. Readyset was designed specifically for this world.

• Drop-in acceleration layer. Readyset sits in the query path and returns results from cache transparently when it’s safe to do so. You don’t rewrite endpoints or wire your app to custom materialized views.
• Deep Caching built for correctness. Our dataflow engine tracks data changes and updates precisely, so cached reads stay fresh across writes, exactly what OLTP workloads demand.
• SQL-aware by design. Readyset’s SQL parsing and normalization detect deterministic, parameterized patterns (common in modern application frameworks) and serve from cache automatically.
• Operational safety. Shadowing, gradual rollout, and instant bypass keep correctness intact while you gain speed.
• Proxy control plane. QueryPilot routes queries to Readyset’s cache layer or to your primary database based on real-time telemetry.

As a reminder, Readyset’s caching layer delivers massive scale while relieving your primary database load, not through magic, but because cached queries simply avoid work. No planning, or disk IO, just a really fast lookup. QueryPilot exists to make this power easy to adopt. It is the bridge and the tooling that helps teams plug in the Readyset cache and see results quickly, without changing application code.

QueryPilot as the enabler

As we open QueryPilot’s MySQL Public Beta today, we're inviting developers and businesses to experience firsthand what zero-touch database performance feels like. We're also opening a Private Preview waitlist for Postgres support, prioritizing teams who are already leveraging or planning to leverage AI-generated application workflows.

The future is clear to me: AI-driven applications will define the next generation of OLTP workloads. At Readyset, we're committed to building the foundational technology necessary to power this future, automatically, safely, and at scale. 

Today, QueryPilot accelerates performance through automatic caching. But its role as a smart proxy puts it in a unique position to do more. In the future, we envision QueryPilot becoming increasingly intelligent, surfacing insights, making routing decisions, and coordinating more advanced optimizations dynamically. Caching is just the beginning.

– Gautam

Get started

• Join the MySQL Public Beta
• Join the Postgres Private Preview waitlist