Enterprise Knowledge Search · Real-time Analytics · Security and Compliance
PayPal wanted thousands of employees to query petabytes of governed finance data conversationally, but it had to avoid plaintext secrets, AI hallucinations, inconsistent metric definitions, and data-loss-prevention risks that are unacceptable in financial services.
Reported impact
PayPal delivered governed self-service analytics to 5,000 users, accelerated insight generation and dashboard creation, and proved that conversational data access can meet the accuracy and security bar required in regulated financial operations.
PayPal layered conversational analytics on top of its existing Google Cloud data stack by combining BigQuery, Looker, and Looker MCP Server inside a tightly controlled enterprise architecture. BigQuery remains the underlying warehouse for petabyte-scale data, while Looker provides the governed semantic layer and LookML business definitions needed to keep metrics consistent. To expose that semantic layer safely to LLMs, PayPal deployed Looker MCP Server into its internal MCP Hub behind an AI proxy layer, then wrapped it with an automated security handshake that fetches credentials from Looker, encrypts them into a secure vault or database, and stores access tokens in Redis rather than on user machines. Claude queries the Looker model instead of raw tables, which lets PayPal deliver natural-language analytics with near-100% accuracy while preserving auditability, row-level governance, and strict InfoSec controls.
Users ask questions in Claude Desktop, which connects through PayPal's AI proxy and MCP Hub to Looker MCP Server. The server retrieves governed context from Looker, Looker generates SQL against BigQuery, and the response returns as a natural-language answer grounded in approved metric definitions rather than raw table guesses.
By grounding AI interactions in Looker's governed data models, PayPal ensures accuracy and compliance while enabling 5,000 employees to query petabytes of data using natural language.
PayPal adopted Looker's Model Context Protocol (MCP) Server to provide secure, conversational analytics to its workforce.
When the LLM receives a prompt, it doesn't query the raw database; it queries the Looker model.
delivers almost 100% accuracy against our defined metrics.
the system pushes them directly into a secured vault/database with encryption, while access tokens are encrypted and stored in Redis cache
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
Google Cloud Sample Modified · Jun 18, 2026 HiPay HiPay needed to replace a legacy BI environment that was causing dashboard fatigue, low self-service adoption, and duplicated calculations, while still preserving strong metric governance and security for payment… drives 50% higher BI adoption BigQuery · Gemini Enterprise Agent Platform · Looker Google Cloud Sample Archived · Apr 13, 2026 Premji Invest As an investor in both private equity and public markets with a mandate to generate returns to fund the philanthropic endeavours of the Azim Premji Foundation, Premji Invest must identify opportunities quickly and… Extends analysis of potential investment targets by 5X through Gemini Enterprise multimodal and multilingual… BigQuery · Cloud Storage · Gemini 
AWS Sample Oct 29, 2024 Vertex Pharmaceuticals After designing an experiment, they manually trained the base ML model about what they hoped to identify from the images, labeling each area of interest by hand. In the legacy system architecture, all scientists shared… 11x accelerated image segmentation AWS Fargate · AWS Lambda · AWS Step Functions 