Databook

This update restores the missing software license information on candidate pages, which had temporarily appeared blank due to a spelling mismatch. It also standardizes the spelling of descriptive text across the platform while ensuring all underlying data structures remain stable and accurate.

This update improves the Software Licenses page by replacing stale, hardcoded statistics with live, computed figures and clarifying active versus expired contracts. It also adds a new CSV export for license capabilities, replaces raw code tags with readable labels, and introduces interactive search, sorting, and pagination for license families.

This update improves transparency on license family pages by clearly labeling whether license classifications were manually reviewed, automatically generated by AI, or a mix of both. These new labels help public users easily distinguish verified, high-confidence data from automated estimates.

This update publishes the Licenses analysis to the public website, adding it to the navigation menu. It refines the data classification of software licenses and support contracts down to the individual product level, resulting in more accurate spending breakdowns, and splits IBM products into distinct functional categories.

This update improves the performance of the public-facing Databook API, specifically speeding up the related-notices lookup on contract pages. By optimizing database queries, the response time for these pages has been reduced from over 100 milliseconds to under 1 millisecond.

This update introduces a web-verification feature for software licenses that automatically validates product identities and requires citations for all findings. It successfully identifies non-software services and vendor portfolios within the data, ensuring higher accuracy and preventing incorrect product merges.