Anomalo Transforms Unstructured Data for Enterprise AI
Anomalo has announced the general availability of its Unstructured Data Monitoring product, now enhanced with a major new feature called Workflows that promises to help enterprises tackle one of their biggest challenges in deploying generative AI applications.
The announcement addresses a critical enterprise pain point: while companies rush to implement AI-powered chatbots and retrieval-augmented generation (RAG) systems, most struggle with the quality and trustworthiness of their unstructured data - the documents, emails, call transcripts, and support tickets that make up roughly 80% of enterprise data stores.
Whether deploying RAG systems or customer-facing chatbots, enterprises need to bring high-quality, domain-specific data to their LLMs. The challenge lies in the unknown. Companies do not know what is in their unstructured data, let alone trust it, making it hard to bring production-ready Gen AI applications to market.
"Everyone's talking about unstructured data for Gen AI but the real breakthrough is solving for both quality and insights within this type of data," said Elliot Shmukler, co-founder and CEO of Anomalo.
The company claims the new Workflows capability transforms what has traditionally been a months-long manual process into a 10-minute automated analysis. The platform can process more than 100,000 documents in a single run and operates continuously as new data arrives, identifying quality issues like duplicates, personally identifiable information (PII), abusive language, and inconsistent formatting.
Beyond quality control, Workflows enables enterprises to convert unstructured content into structured data ready for downstream analytics and AI applications - a crucial step for companies building domain-specific AI systems that require high-quality training data.
Anomalo claims to be the first company to announce AI-powered monitoring of unstructured text, having introduced the capability in June 2024 before adding additional features in November. The company's approach differs from traditional rules-based systems by using machine learning to automatically detect data quality issues across both structured and unstructured datasets.
The platform offers 15 out-of-the-box quality checks while allowing customers to create custom quality criteria tailored to their specific use cases. This flexibility appears designed to address the varying needs of enterprises across different industries - from retail companies mining support tickets to understand customer dissatisfaction to restaurant chains analysing guest feedback.
The general availability of Unstructured Data Monitoring with Workflows represents Anomalo's attempt to extend its data quality expertise into the rapidly growing market for enterprise AI infrastructure, where the quality of training and retrieval data often determines the success or failure of AI applications.