Relyance AI’s new data tracking tool paves the way for greater AI accountability and explainability

The ambitious big-data governance startup Relyance AI Inc. wants to tackle one of the biggest roadblocks preventing highly regulated enterprises from adopting artificial intelligence with the launch of a new platform called Data Journeys.

The company said today that Data Journeys is all about helping companies to understand how their data travels across a labyrinth of different computer systems within their organization. It keeps track of where the data resides, where it came from, where it’s going next, and also how and why it’s being used by different applications and services.

Relyance AI is a five-year-old startup that came to prominence last October when it raised $32 million in a Series B funding. It’s the creator of a governance platform that aims to provide clear visibility into enterprise-wide data, as well as controls to safeguard that information.

Its platform works by scanning all of the data within an organization, including everything that resides in their applications, databases, AI models and code repositories. It then compares this data with all of the policies stated within the company’s contracts and regulations to ensure nothing is amiss.

With Data Journeys, Relyance AI is now using the same core technology to keep track of data throughout its entire journey as it flows to different systems, applications and AI models.

The startup says Data Journeys is a big improvement on traditional data lineage tools that are used by enterprises to track data. Those systems are restricted to only tracking data movement on a table-to-table or column-to-column basis, but they lack wider visibility. For instance, if data is moved from an Amazon S3 bucket to a completely new database, it loses track, so no one can say where it originally came from. That means there’s no clarity regarding the transformations that took place as the data moves from third-party systems, application programming interfaces and retrieval-augmented generation architectures.

With Data Journeys, Relyance AI wants to provide a much more comprehensive view. The aim is to reveal the full data lifecycle, from where it’s collected, where it’s moved and how it’s transformed along the way, and also visibility into how it’s being used. It does this by using continuous source code analysis, giving it greater context around how the data is being processed, compared to systems that only plug into data repositories.

The startup says enterprises need this enhanced visibility because they’re under increased pressure from regulators. It notes that more than 25% of the Fortune 500 firms have labeled AI regulation as a risk in filings with the U.S. Securities and Exchange Commission.

Enterprises are eager to embrace the possibilities of AI, but doing so comes with an urgent need to ensure accountability over how such systems use their data. In an interview with VentureBeat, Relyance AI co-founder and Chief Executive Abhi Sharma said the status quo of AI being seen as some kind of “black box” of data, sources and queries must change.

“We’re providing transparency not as a feature, but as a necessity for fairness, accountability and trust for our customers,” he explained.

The Data Journeys offering is particularly interesting for companies in the tightly-regulated healthcare industry.

“After seeing Relyance AI Data Journeys, we immediately recognized its potential to revolutionize our approach to responsible AI development,” said CHG Healthcare Inc. Privacy Officer Heather Allen. “The automated, context-aware data lineage capabilities would address our most pressing challenges. It represents exactly what we’ve been looking for to support our global AI governance framework.”

According to Sharma, Data Journeys can solve four major impediments to enterprise AI adoption: risk management, precise bias detection, explainability and accountability, and regulatory compliance.

For instance, Data Journeys’ tracking capabilities enable potential bias in AI models to be tracked directly to the source. Sharma explained that in many cases, bias is not from poor quality data in the underlying dataset, but rather the journey that information took.

Moreover, the enhanced explainability will be vital for AI-powered decision-making in areas such as loan approvals and medical diagnoses, as it will allow models to show why they came to a certain conclusion.

“Many times, the incorrect behavior of the model is completely dependent on the multiple steps it took before the inference time,” Sharma said.

Looking forward, Sharma said his company is looking to build a “unified AI-native platform” for data governance, management and compliance, and he believes Data Journeys will be a critical component. “AI agents are going to run the world, and we want to be that company that provides the infrastructure for organizations to trust and govern it,” he said.

Image: SiliconANGLE/Meta AI