Tray.ai Launches Data Engineering Capability to Help Break Down Failure Rates for AI Projects

Tray.ai today announced the general availability of its Data Engineering capability as part of its unified automation platform. The new offering is designed to help companies eliminate significant roadblocks to successful AI implementations with siloed data pipelines and disconnected integrations tools in the crosshairs.

The release features capabilities for bringing together in-flight data preparation and agent development. With Tray’s Data Engineering functionality, organizations can develop agents to transform data into clean, usable information in context — preparing data as part of automation workflows from ingestion all the way to intelligence.

Enterprise buyers have demanded modernization of data stacks for AI. Traditional automation platforms were not built to ingest data and power AI use cases at scale. Instead, enterprise teams have often been required to manage separate preprocessing stacks, tools and scripts outside of agent workflows. This disconnect can create bottlenecks when scaling AI pilots to production.

“We see companies look to modernize their data stacks for AI, but don’t want to replace their entire infrastructure with another siloed tool,” said Julia Adams, chief strategy officer at Tray.ai. “Developers should be able to modernize their data transformation workflows and prepare information for AI agents without having to manage new infrastructure.”

Business adoption of generative AI at scale depends on AI-ready data. Inconsistent, inaccessible, and dirty data can derail machine learning model accuracy and agent performance leading to elevated levels of technical debt. Research commissioned by Tray.ai from analyst firm Gartner found AI projects have greater than a 60 percent chance of failing to deliver on business service-level expectations if an organization lacks AI-ready data practices in place.

The global database market generated revenue of over $33 billion in 2023. However, as APIs continue to mature and industries continue their rapid shift to software-driven automation, data movement modernization has emerged as a new battleground for platforms looking to capitalize on AI at scale.

At the foundation of Data Engineering is the new Tray SQL Transformer, an embedded virtual database that allows developers to run ANSI SQL against their data while it’s in flight through automation processes. The feature empowers organizations to reshape, join, and transform their data alongside their automation development workflows. Leading tech companies like Microsoft, Amazon, and Google all offer some form of serverless data lake functionality built on SQL. Making SQL available at scale for transformation needs has been a highly requested feature from Tray.ai customers, according to the company. 

Developers can also load files using Parquet and JSON file types into Tray Workflows. In addition to joining and transforming datasets in-flight, Tray will apply ANSI SQL transformations to the data without requiring users to manage additional infrastructure or database layers. Tray says developers will be able to apply transformations to millions of records in a single operation.