Tradeshift Moves to Agentic AI with Amazon Quick, Speeding Up Queries 30 Times
Amazon Web Services
Tradeshift replaced its outdated internal BI tool with Amazon Quick, an agentic AI solution built on AWS. The deployment cut query execution time by up to 30 times, reduced total cost of ownership by 40%, and turned embedded analytics into a revenue-generating product.
Tradeshift, a platform for accounts payable management and electronic invoicing, originally used its own BI tool that struggled with growing data volumes and customer expectations. It supported a maximum of 10,000 rows per query, a 25 MB report limit, and stored only six months of history. The company migrated to Amazon QuickSight, an AI-powered analytics space that includes a chat agent for natural language queries, automated workflows (Flows), and a research module (Research). Results: query response time dropped from 45–90 seconds to under 3 seconds, total cost of ownership decreased by 40%, and infrastructure costs fell by 35%. Internal teams saved 8.5 hours per week on manual reporting, while external buyers saved 6–8 hours per user. The company launched embedded analytics with two tiers: Standard (standard dashboards) and Premium (custom dashboard creation, scenario modeling, and access to agentic AI). In the first year, 50% of enterprise buyers actively use the platform, and customer retention with analytics is 10% higher. The architecture includes 16 embedded dashboards, over 270 SPICE datasets, and the AP Auditor chat agent connected to 11 reference documents, 9 dashboards, 14 query topics, and 68 tools via the Model Context Protocol. Security is ensured with Okta SSO, custom namespaces, signed URLs, and row-level security with 14,000 rules.
Source: AWS ML blog —
original
