Auspex

Semi-structured Data Extraction

Humans generate a huge amount of semi-structured data in spreadsheets, documents, presentations and emails. This data evolves over time as new data is added: columns, rows and paragraphs changing the structure of the data. Traditional data extraction techniques struggle with this change, as well as the renaming, typos and slang which often creep in.

Auspex has been built with these challenges in mind. The toolset has two sides, the first is AI enhanced discovery, which can ingest and assess documents, finding the patterns and identifying the key data points. Verified by a human the discovery process creates a standard pattern which it passes to the second side of Auspex, execution. Taking the pattern Auspex can automatically, and deterministically ingest data of the same semi-structure (e.g. monthly reports, or weekly status updates). If the execution step encounters change it can pass it automatically back to the first process, to identify possible causes and solutions.