Dinoticia is supporting the automation of unstructured document preprocessing, a problem cited as a major cause of early failure in corporate AI projects. On Tuesday, the company said it officially launched its Seahorse Cloud SaaS service, enabling users to handle the entire process in a single environment, from document upload to parsing, structuring and vectorisation.

To apply generative AI and AI agents to work, unstructured data such as PDFs, images and documents must first be converted into a form that AI can process. But unstructured data in different formats required a separate data pipeline or manual preprocessing.

The service is a managed offering that integrates vector database-based RAGOps, or retrieval-augmented generation operations, and AgentOps, an AI agent operations function. Companies can perform vector data processing, RAG configuration, and agent design and operations in one environment without building separate infrastructure.

For document parsing, it applies VLM, or vision-language model-based, layout analysis technology to distinguish page structure and table and image areas. It then uses OCR, or optical character recognition, and LLM, or large language model-based, text refinement to structure documents by semantic unit. It detects and restores tables separately to minimise information loss and improve question-and-answer accuracy. It also supports text conversion for image-based documents such as flowcharts so AI agents can use them for contextual search.

It is provided in an Amazon Web Services environment and can be used immediately through the official website and control console. It allows companies to adopt the service while keeping their existing cloud infrastructure. It provides trial credits to early users.

A Dinoticia official said, "Companies can process countless unstructured documents in a single integrated SaaS environment and derive data-based intelligence."