Scaling AVs in 2025: Why Smart Data is the Missing Piece

From Pilots to Production: What's Holding Back AV Scalability?

Autonomous vehicles (AVs) have come a long way since 2023, but as of mid-2025, they still haven’t reached mainstream deployment.

In our latest webinar, Heex VP of Business Development Pierre Maillot sat down with Marek Vanzura, Chief Innovation Officer at autinno, to discuss the technical and operational challenges blocking large-scale AV deployment.

Marek, who has tested dozens of AV platforms, shared candid insights into what’s working, what isn’t, and why smart data management could be the key to unlocking global scalability.

The Hidden Bottlenecks Slowing AV Progress

While public perception and regulation are often cited, Marek pointed to deeper, structural issues.

Complex and fragmented tech stacks

Most AV platforms rely on a patchwork of tools and components, making them hard to maintain or scale.

Siloed data and workflows

QA teams, developers, and operations often lack shared visibility into what’s happening on the road.

Overwhelming data volumes

Some vehicles generate multiple terabytes per day, and most of it is either discarded or never touched again.

"We’re not short on data, we’re short on clarity," Marek noted.

"Debugging one issue can mean hours of digging through raw logs."

This data overload doesn’t just slow development. It blocks real-world deployment. Without fast feedback loops and cross-team alignment, AVs stay stuck in pilot mode.

Why Big Data Alone Isn’t Enough

Intel once estimated that AVs can produce up to 4 or 5 terabytes of data per hour. In practice, Marek confirmed that the challenge isn’t hypothetical. It’s real, and it’s growing.

Teams often struggle with:

• Unstructured data pipelines
• Poor traceability of issues
• Limited visibility for non-engineers

In many deployments, the bottleneck isn’t perception or control. It’s data. How to capture it, manage it, and extract value without being buried under noise.

Turning Big Data into Smart Data: A Better Approach

That’s where Smart Data comes in.

As Heex CEO Bruno Mendes Da Silva explained during the session, the Heex platform is designed to address this exact challenge.

Edge agent → Installed directly on the robot or vehicle

Event-triggered data capture → Only records what matters

Centralized cloud portal → Accessible to QA, Dev, and Ops

OTA deployment → Easy to scale across fleets

"We don’t need more data," Bruno said. "We need better data — data that’s meaningful, contextualized, and available when and where it’s needed."

Marek agreed that this kind of solution could significantly shorten the time from test to validation, especially in multi-vehicle environments.

For more details about our product you can access our product page or our documentation.

What’s Next for the AV Industry?

As the discussion wrapped up, Marek emphasized the importance of:

• Standardized data workflows
• Better cross-team visibility
• Smart, explainable tools to support scale

Scaling AVs isn’t just about autonomy. It’s about making sense of the massive data they generate.

Want to dive deeper into the conversation? Watch the full replay