The Future Is Unstructured: How Anomalo and Databricks Are Building the Data Foundations for an AI World

As AI practitioners and data executives paint San Francisco red this week for the Databricks Data + AI Summit (DAIS), the spotlight is squarely on AI. But here’s the hard truth: AI projects fall short when data teams rely on data they do not understand or are riddled with data quality issues. This is especially true of unstructured data, which makes up over 80% of enterprise data today.

That’s where Anomalo’s new Unstructured Data Monitoring product comes in. Anomalo integrates directly with Databricks to turn raw unstructured content into compliant, AI-ready data that is governed at scale. Paired with Databricks, Unstructured Data Monitoring provides the quality foundation your AI initiatives need—so you can build smarter, safer, and faster. 

Why Unstructured Data Needs Rethinking

Databricks gives you the foundation to build powerful AI apps — from flexible data storage to Llama and Claude to scalable compute and developer tools like AI Builder, Mosaic, and MLflow. Even the most advanced AI systems will falter if they rely on ungoverned, low-quality data as context for decision-making. This is especially true for unstructured data—PDFs, emails, documents, reports, contracts, web pages, screenshots. It’s messy, inconsistent, and often invisible to traditional governance. When agents act on the wrong signals, outcomes can be unreliable at best—and risky at worst.

And until now, this data has been underutilized and sometimes invisible.

Enter Anomalo: Automated Trust for Unstructured Data

With Anomalo’s Unstructured Data Monitoring product, enterprises can curate unstructured text documents and evaluate them for data quality around various document and document collection characteristics, including document length, duplicates, inconsistencies,  topics, tone, abusive language, PII, and sentiment. Customers can quickly assess the quality of a document collection, dramatically reducing the time needed to curate, profile, and leverage high-value unstructured text data. In addition to Anomalo’s 15 out-of-the-box issues, customers can create their own custom issues to look for and designate what classifies as high or low quality for their documents with custom severity scores.

But what good is quality if you don’t understand the data itself? Anomalo’s Unstructured Data Monitoring also lets enterprises extract insights from the vast volumes of unstructured data stored in Databricks. A key feature of the product is Anomalo Workflows, a hub for managing and monitoring unstructured data. This moves the Anomalo product beyond just being a platform for data quality. With Workflows, customers can: 

1. Identify and correct quality issues like duplicates, errors, PII, and abusive language
2. Analyze large volumes of unstructured content to uncover patterns and extract meaningful insights
3. Convert unstructured content into structured data ready for downstream analytics and AI workflows
4. Curate document collections into clean, reusable sets for training or retrieval

Our mutual customers with Databricks love this.

“In the restaurant service industry, understanding and acting on guest experiences is critical—and that means unlocking insights from the tens of thousands of unstructured comments we receive each month. Through our collaboration with Anomalo, we’ve started exploring how their Unstructured Data Monitoring can surface meaningful patterns in support tickets and guest feedback. We’re excited about the power to turn this data into actionable insights, strengthen our GenAI initiatives, and bring high-quality unstructured data into everything we build.”

– Sid Stephens, Data Governance Leader and Databricks customer

Build AI with Confidence, with Databricks and Anomalo 

With Anomalo, you can monitor document collections directly from Databricks. We analyze this data using Databricks-hosted LLMs or externally-hosted models to classify content, redact sensitive information, and detect anomalies that threaten quality and trust. This means you can use the models already approved with your organization. The immediate result is enterprise-scale curated, compliant datasets ready for RAG Pipelines, all within your governed, Unity Catalog-controlled environment.

“As enterprises scale AI on Databricks, their most valuable insights are increasingly buried in their own unstructured data — documents, transcripts, logs, and more. But unlocking that value requires accessibility and trust. That’s where Anomalo comes in: they automatically detect and fix quality issues, curate and enrich unstructured data, and make it instantly usable on Databricks with a simple prompt. It’s more than monitoring — it’s an intelligence layer that transforms messy data into AI-ready fuel.”

– Ari Kaplan, Head Evangelist and AI Expert at Databricks 

Here’s how that works with Databricks and Anomalo in 4 easy steps.

1. Raw, unstructured documents get analyzed and redacted by Anomalo. These are passed through the Mosaic AI Gateway to leverage Databricks-hosted or external LLMs. 
2. Anomalo identifies issues with unstructured data, such as % of documents containing PII or documents contradicting one another.
3. Anomalo also uncovers insights, such as the most common issues in a collection of customer support calls. 
4. Clean documents are written to a  Unity Catalog Volume  to ensure governance, access control, and data lineage
5. Data is now ready for Databricks AI Builder for drag-and-drop app development

Anomalo integrates with Databricks to turn raw unstructured content into compliant, AI-ready data: governed, compliant, and monitored at scale. 

If you’re a data analyst, governance leader, or ML practitioner building AI solutions on Databricks, this is your moment to operationalize trust in unstructured data. Request a demo to learn how Anomalo can help your enterprise unlock generative AI assets with confidence.