Ataccama ONE brings document AI to Snowflake for better data
Ataccama has announced the integration of its unified data trust platform, Ataccama ONE, with Document AI on the Snowflake Marketplace, allowing businesses to transform unstructured content into structured data for analytics and artificial intelligence applications.
The announcement from Ataccama enables enterprises to extract, structure, govern, and monitor the quality of unstructured data, making a greater proportion of their information usable for analytics, AI, and business operations. With the majority of enterprise information now classed as unstructured data and continuing to grow rapidly, many organisations struggle to manage this resource effectively.
Industry research from IDC indicates that unstructured data accounts for most enterprise data and is expanding by more than 55% annually. According to market findings, 95% of businesses face difficulties in managing their unstructured data, with over half citing it as the most challenging type of information to govern. Unstructured data frequently remains siloed and difficult to leverage, creating operational risk and impacting the reliability of AI systems as these systems increasingly rely on such data for powering large language models and retrieval-augmented generation applications.
Through the integration of Ataccama ONE and Document AI within the Snowflake environment, enterprises can convert documents such as contracts, invoices, and PDFs into structured records. Natural language prompts—such as, "What is the effective date of the contract?"—are processed by the Arctic-TILT large language model developed by Snowflake, generating structured outputs stored directly in Snowflake tables.
Ataccama ONE then connects to the resulting data tables to profile the data, perform quality checks, and manage governance policies on the structured outputs. The system also allows companies to follow the data through analytics, reporting, and AI workflows by capturing lineage at the table level. Additional metadata can be added from the original documents to increase traceability where required. This process aims to reduce manual intervention, build trust in the data, and enable repeatable workflows for business teams.
Speaking to the importance of unlocking value from unstructured data, Sam Wong, Senior Director of Data & AI at a global beverage company, said: "Unstructured data is an untapped data source as real business context lives there, but it's also the hardest to govern. Documents, contracts, and communications contain the terms, conditions, and risks that structured systems miss. Without a way to extract, validate, and manage that information at scale, AI lacks the foundation it needs to be reliable. With Ataccama ONE and Document AI inside Snowflake, organizations can turn thousands of documents into trusted, structured data. That will give us improved analytics, enhanced data quality, and a better foundation for powerful and trustworthy AI."
This integration offers several capabilities for users. Companies can extract structured data from documents using natural language, specifying prompts such as "What is the payment term?" to convert unstructured information into structured outputs without the need for custom code. The extracted data is immediately available for use in reporting, analytics, and AI activities, as it is saved directly into Snowflake tables, ready for consumption by business intelligence tools and model pipelines without further transformation.
Ataccama ONE provides automated profiling and rule-based validation to monitor the quality of unstructured data on a continual basis, assisting teams in detecting inconsistencies and managing risks early. Document AI models can be trained and reused within Snowsight, allowing for standardised extraction across various document types at scale, including contracts, invoices, and policies. All processes, validation, and governance are performed natively in Snowflake, reducing integration complexity and improving security.
Jay Limburn, Chief Product Officer at Ataccama, said: "Unstructured data remains a black box for most organizations, even as it becomes critical for AI and business operations. Without a way to structure, govern, and trust that information, enterprises risk missing the full value of their data. Ataccama ONE combines data quality, governance, observability, lineage, and master data management in a single platform and now extends those capabilities to unstructured content. This allows organizations to improve trust and confidence in all their data, structured and unstructured alike, and build a stronger foundation for AI, analytics, and operational decision-making."
Kieran Kennedy, Vice President, Data Cloud Products at Snowflake, said: "Ataccama's presence on Snowflake Marketplace reinforces the value of our integrated platform approach that allows our partners to bring their innovative solutions to market within the Snowflake environment. With this solution, joint customers have the power to streamline document extraction, ensure data quality, and accelerate insight delivery, all within a governed and scalable environment."
