Komprise launches Transparent File Tables for AI analytics

Komprise has launched Transparent File Tables, a product that gives AI and analytics platforms a structured view of enterprise unstructured data.

The software exposes unstructured data as an Apache Iceberg table for use in platforms including Snowflake and Databricks.

Unstructured data makes up more than 80% of the typical enterprise data footprint, according to IDC, yet less than 1% is used in AI. That gap has persisted because much of the data lacks a consistent schema, often has quality issues, and is difficult to transfer at scale across network-attached storage and cloud systems.

Komprise is positioning the offering as a way for data engineers, data scientists, and analysts to work with file-based data in existing analytics environments without first copying large volumes into a data lakehouse. Instead, the software presents metadata in tabular form and points users to the underlying files.

How It Works

The system builds on Komprise's existing architecture for indexing enterprise data across datacentres and hybrid cloud storage into what it calls a Global Metadatabase. Users can enrich files with additional context through content, header, and sensitive data scanning, as well as metadata tagging, before exporting a table for use in analytics tools.

Enterprise data teams can then query those tables in Apache Iceberg using their preferred business intelligence and analytics tools. Governance is applied through user access permissions.

If a workflow needs the full file rather than metadata alone, the software uses Komprise's Transparent Move Technology to fetch data only when needed instead of moving an entire dataset in advance. That allows remote data to remain in place until an AI or analytics process requires access.

The approach addresses one of the central obstacles in enterprise AI projects involving file and object stores: the cost and complexity of moving petabytes of data across different storage systems and locations. In many organisations, ingesting large unstructured datasets into analytics platforms can take weeks or months.

Target Uses

Komprise highlighted uses in analytics and AI pipelines that combine structured and unstructured information. In one example, a pharmaceutical data analyst could query project files produced by instruments and laboratories, then join that information with financial data from enterprise resource planning systems and instrument data from Benchling in a single analytics environment.

Another example involved a media and entertainment AI workflow, in which an AI agent could use structured project data to identify relevant media archives and then combine that with Transparent File Tables to narrow the set of scripts selected for summarisation.

The launch reflects a wider effort across the data industry to make unstructured data more accessible to AI systems without forcing companies into large migration projects. Vendors have increasingly focused on metadata, cataloguing, and open table formats as a bridge between traditional storage estates and newer analytics platforms.

Komprise said its software can classify unstructured data globally and generate a higher-quality schema for downstream use. The aim is to help data teams discover relevant content, apply context, and work with that data in familiar query environments.

Kumar K. Goswami, Chief Executive Officer and Co-Founder of Komprise, outlined the company's case for the product. "The reason 99% of enterprise unstructured data has been dark to AI and analytics is because discovering and generating its schema and moving it is inherently complex and costly," he said.

He added: "Komprise brings to light the huge petabytes of enterprise unstructured data in a form that data teams can access easily and transparently for analytics. Komprise Transparent File Tables opens a whole new world to AI."

The product is available through an early access programme.