Customer Identification via Email Content Matching (Data App)

<p><strong>Fault-Tolerant Customer Lookup from Unstructured Data</strong></p><p>This Data App demonstrates how the <strong>M|Box Table Indexing</strong> nodes can be combined with LLMs to find records in large datasets using unstructured text data.</p><p>The Customer Lookup Data App:</p><ul><li><p>Loads a dataset containing 100,000 customer records, including names, phone numbers, addresses, and other attributes.</p></li><li><p>Builds a multi-field search index using the <strong>Table Indexer</strong> node to enable efficient and robust lookup across multiple customer attributes.</p></li><li><p>Uses the LLM-powered <strong>Document Entity Extractor</strong> component to extract relevant customer metadata from a user-provided email.</p></li><li><p>Matches the extracted information against the indexed dataset using the <strong>Table Index Matcher</strong> node to reliably identify the correct customer record.</p></li></ul><p>M|Box's fast, fault-tolerant indexing enables reliable customer identification, even when the input contains typos, incomplete information, or minor inconsistencies introduced by user input or LLM-generated metadata.</p>