We have big and exciting news: Databricks has acquired Quotient.
‍

‍When we started Quotient in 2023, we had a vision. We believed the biggest impact of AI on real problems wouldn’t come from ever-larger foundational models alone. It would come from deploying models specialized for specific domains, fine-tuned on relevant data, and optimized for concrete workflows inside real-world agents. To make that possible, we saw the need for custom evaluation technology that went beyond generic benchmarks.

Early AI products like ChatGPT and GitHub Copilot had captured the world’s imagination, but the gap between benchmark performance and true impact in the field was impossible to ignore. We’ve spent the past few years building infrastructure that enables developers to measure and systematically close that gap in their own domains. Databricks gives us the platform, scale, and reach to push that work much further.

What we built, and what we learned

We began with evaluation and monitoring, building the observability layer teams needed to understand what their AI systems were actually doing in production.

We initially focused on developer experience and high-signal evaluators for the failures that matter most: hallucinations, policy violations, and broken tool use. Over the past year, we’ve been working on technology that transforms production telemetry into structured reward signals and post-training pipelines. This allowed us to create agents that automatically learn on the job and build domain expertise over time.

We learned the most from working with customers who operate complex, production-grade AI systems. For these organizations, reliability is non-negotiable. Our platform evolved to meet their needs.

Why Databricks

We believe that our technology is ready to close the gap for domain specific AI. To make this happen, we need distribution, to be co-located with customer data, and the ability to train and deploy custom agents. Databricks has the world’s best platform for all of this.

This also feels personally meaningful. Our work on GitHub Copilot — the first AI coding system deployed at massive scale — shaped how we fundamentally think about building AI coding agents. At Databricks, we see that same frontier emerging again with products like Genie, Genie Code, and Agent Bricks, bringing agentic workflows into core data use cases and workflows. 

We’re excited to help push that boundary forward.

You can read more in Databricks’ blog post about the acquisition.

Thank you!

Building Quotient with this community ranks among the most energizing experiences of our careers. You validated the mission, pushed the product to be better, and kept us honest about what actually matters when shipping AI in the real world.

We think you’ll like what we’re going to build at Databricks.

Onward!

— Julia, Freddie, and the Quotient Team