Today, we’re introducing definity’s Agentic Data Engineering platform – alongside a $12M Series A led by GreatPoint Ventures, with participation from Dynatrace, StageOne Ventures, and Hyde Park Venture Partners.

This milestone marks an important step in how enterprise data platforms are operated.

The Operational Problem in Modern Data Platforms

Over the past few years, we’ve worked closely with enterprise data teams running some of the most complex lakehouse and Spark environments.

Despite the rapid evolution of the data stack, one pattern has remained consistent: operating Spark data pipelines at scale is still too manual, too reactive, and too expensive.

When pipelines break, performance degrades, or costs spike, data engineers are forced into a familiar cycle – spending days digging through logs, correlating dashboards, and trying to understand what happened after the fact. Even routine optimization work can take days of investigation.

This is not a tooling gap. It’s an operating model gap.

Lakehouse platforms were built to run pipelines – but not to proactively monitor, mitigate, protect, fix, optimize, and evolve them in production.

Observability helped bridge part of that gap by providing visibility. But visibility alone doesn’t enable control. It doesn’t allow teams to intervene, optimize, or prevent issues proactively.

As a result, data engineering teams remain reactive, even as the systems they manage become increasingly critical to the business.

Why AI Agents Haven’t Transformed Data Engineering (Yet)

At the same time, AI agents are reshaping how systems are designed and operated across the enterprise.

But in data engineering, their impact has been limited.

The reason is simple: agents today lack the foundation required to operate production data systems effectively and reliably.

Without full-stack runtime context (across data behavior, pipeline execution, system configuration, and infrastructure performance), the ability to control execution in-motion, and native guardrails to ensure safe operation – agents are confined to analysis and recommendations.

They can assist. They can suggest. But they cannot operate.

Introducing Agentic Data Engineering

This is the gap definity was built to solve.

We are introducing a new operating model for data platforms: Agentic Data Engineering.

And it requires a new foundational layer in the data stack.

At its core, definity provides the runtime infrastructure that allows AI agents to understand, optimize, and take action across data pipelines – in real time.

Instead of analyzing data after the fact or relying on historical signals, definity operates directly within pipelines during execution. By capturing unified, in-motion context across infrastructure, pipeline logic, and data, the platform enables continuous optimization, proactive mitigation, and reliable autonomous operation.

This shifts data engineering from reactive monitoring to active, intelligent operation.

Not dashboards. Not alerts. Not recommendations.

Actual operation.

This is not a theoretical shift. Enterprise teams are already using definity in production to:

• Reduce platform costs through job-level optimization
• Resolve complex pipeline issues significantly faster
• Prevent incidents before they impact downstream systems and business operations

And as platforms scale, these capabilities move from “nice to have” to essential.

Building the Runtime Layer for Data Engineering

The next evolution of the data platform is not another layer of monitoring or analytics.

It’s a runtime layer for operation – that enables AI agents (and humans) to move beyond visibility and into continuous, intelligent, and reliable control of production systems.

That’s what we’re building at definity.

Looking Ahead

We’re grateful to the customers and partners who have been building with us from the early days, shaping both the product and the vision.

We’re thankful to GreatPoint Ventures, Dynatrace, StageOne Ventures, and Hyde Park Venture Partners for their continued support and conviction.

This is still early – but the shift from monitoring to operation is already underway.

If you’re operating large-scale lakehouse or Spark environments and thinking about what comes next, we’d love to connect.

‍

Read more on VentureBeat or the full press release