The Formatting
CrushBank’s proprietary ingestion approach makes getting data from any system of record into your data lakehouse simple.
The CrushBank data lakehouse architecture then stores that data in three formats; unstructured, structured and vectored.
CrushBank’s orchestration agent takes incoming requests and determines which search tools and which data format will best handle that request; vectored for information retrieval or structured for analytics and computational questions. Plus the complete (unstructured) version is always available for review, or a deeper dive.
CrushBank’s architecture normalizes and standardizes all inputs so users don’t need to know what source system they need to access to find the information they need – even if it originally lived across multiple platforms.
CrushBank honors source system permissions, and leveragers IBM’s proprietary AI governance, and the resulting transparency allows for “trust but verify” results.
The Outcomes
CrushBank autonomous agents run off the data set in the data lakehouse. This allows for immediate action based on the previous interactions and tickets. Ticket Auditing, Ticket Agreement Auditor and Ticket Bundler.
Automations leverage previous ticket data to provide alignment, budgets and resource recommendations for client requests. The Classification, Prioritization, Budget and Summarization of Tickets are all automated.
Asking questions with normalized data allows users to use a single format to query data for almost any type of question; how many… where is the… what is the…. These questions are all answerable from the data lakehouse.
Internet Search
- Search the internet for questions you have that are not in your data lakehouse.
- AI-driven search transform how users can find information quick and easy.
- Instead of searching for keywords, users can ask questions naturally and receive synthesized, context-aware insights drawn from vast data sources.