Case Study: TELUS scales 300,000+ AI copilots with turbopuffer

TELUS powers 300,000+ AI copilots with turbopuffer

Telus, a large Canadian telecommunications and technology company, needed a vector database capable of supporting hundreds of thousands of indexes for its new generative AI platform without incurring an exorbitant cost. After five other vector databases failed in production due to scaling limitations and high expenses, they turned to Turbopuffer for a solution.

Turbopuffer provided a massively multi-tenant architecture that builds indexes on cheap object storage, allowing Telus to create a unique namespace for each AI copilot. This solution delivered sub-100ms latency and was dramatically less expensive than alternatives. The implementation was a success, enabling Telus to launch its Fuel iX platform, which is used by over 57,000 employees and has reportedly generated over $90 million in benefits.

Telus