Tray.ai this week added a data engineering capability to its core integrated platform-as-a-service (iPaaS) environment for integrating applications and workflows.

The Tray Data Engineering extension is based on Tray SQL Transformer, a tool the company developed to enable IT teams to execute data preparation tasks directly within a Tray workflow that, via a virtual database, makes it possible to reshape, join, and transform data in flight.

Tray.ai CEO Rich Waldron said that, as the number of applications that incorporate artificial intelligence (AI) agents continues to rise, it’s becoming apparent that there is now a greater need to converge data engineering and application integration within one platform. Ultimately, the goal is to make it simpler for IT teams to get the right data to the right place more reliably using a set of familiar SQL tools, he added.

SQL Transformer makes it simpler to create data pipelines to ingest millions of records that can be joined across multiple files without IT teams having to create scripts or rely on bolt-on tools to pre-process data, noted Waldron. That approach makes it much easier to deduplicate records, standardize text casing, and validate formats before the data ever enters a data warehouse or database, he noted.

Data can be output as JSON objects that are accessed via application programming interfaces (APIs) and AI agents, or written to files for bulk loading into data lakes and data warehouses such as Snowflake, Databricks, BigQuery and Redshift.

Based on a serverless architecture, the Tray platform also now includes support for JSONata Inline Functions for handling complex data transformations across any data type directly within Tray’s visual workflows. By combining native SQL transformation and in-flight data cleansing into a single unified platform, the Tray platform can now be used to consolidate fragmented toolchains to reduce total costs, noted Waldron.

Those capabilities ensure that high-quality data is being loaded into platforms that are being accessed by either agents created on the Tray platform or third-party AI agents that might otherwise generate less reliable output by relying on flawed data that cause them to hallucinate, he added. Data engineering, along with data hygiene, has emerged as the single biggest obstacle to operationalizing AI, noted Waldron.

It’s not clear at what pace organizations are revamping data engineering in the age of AI, but the Futurum Group projects the global data intelligence, analytics, and infrastructure (DIAI) market will grow at a 17% compound annual growth rate through 2028 off a base of $541.1 billion in 2026 to exceed $1.2 trillion by 2031. As those investments are made, the role of IT teams is expected to evolve as manual data engineering workflows as more IT teams automate extract, transform and load (ETL) pipelines, also known as Zero-ETL. In effect, data engineering teams are evolving into shepherds of data that is increasingly being used to drive AI applications and agents.

It’s not clear how long it might take before IT teams make that transition but the one thing that is certain is that as more investments in AI are made the many existing flaws that exist in how IT teams have historically managed data are being increasingly exposed.