Case Study: Gradient Labs builds reliable AI customer support agents with Temporal

Gradient Labs uses AI agents to resolve complex customer issues in financial services

Gradient Labs, a London-based startup building AI agents for financial services customer operations, needed a reliable way to handle the complexity of agentic workflows. Its team wanted to focus on prompt engineering, evaluations, and improving the AI agent itself, rather than building retry logic, error handling, and other scaffolding around high-stakes customer support tasks. They chose Temporal Technologies and Temporal Cloud to support their Go-based AI agent architecture.

With Temporal Technologies, Gradient Labs orchestrates customer service, evaluation, and knowledge-base editing workflows so its agent, Otto, can reliably manage live chat, email, and ticket conversations. Temporal helps the team retry failed LLM calls, maintain state across hours or days, and run durable API actions like account upgrades and payment checks. The result is faster workflow implementation, easier debugging, and higher reliability—helping Gradient Labs’ agents resolve 40–60% of cases out of the box and up to 80% with optimization, while also achieving stronger customer satisfaction than human agents.

Gradient Labs

Neal Lathia

Co-founder and CTO