Customer Experience Agents in Production
Get an inside look at how Lyft, Fastweb + Vodafone, and LATAM Airlines are building customer experience agents that improve through real production use.
How Customer Experience Teams Are Building Agents That Get Better Over Time
Customer experience has become one of the fastest-moving areas for agents, where faster responses, fewer escalations, and more successful resolutions can directly impact conversion, support costs, and customer retention. This guide breaks down how Lyft, Fastweb + Vodafone, and LATAM Airlines are building production agent systems, from customer-facing support agents to internal copilots and platforms that turn conversations into business intelligence.
What You'll Learn
- How Lyft built a self-serve platform that enables operations teams and product managers to launch support agents, supported by realistic simulations, behavior-specific evals, and continuous production feedback.
- How Fastweb + Vodafone use Super TOBi and Super Agent to resolve customer requests, guide call-center consultants, and continuously improve performance through daily monitoring and evaluation.
- How LATAM Airlines used trace-level observability to uncover unmet passenger needs, reduce out-of-scope requests from 13% to 1%, and turn customer conversations into structured knowledge.
- Why structured prompts, production observability, thoughtful agent architecture, and continuous feedback loops are becoming essential for building CX agents that improve over time.
Download the guide to see what production agent development looks like inside customer experience teams operating at scale, and how leading organizations are turning every interaction into an opportunity to improve.
LangSmith powers top enterprise engineering teams
.svg)
.svg)
.svg)
%201.svg){kind=logo}
.svg)
Get Started with LangSmith
See how LangSmith helps you debug, evaluate, and improve your agents in production. Talk to our team or start exploring today.
Related Content
The Agent Development Lifecycle (ADLC)
How the best organizations ship agents repeatedly, safely, and systematically — by building a Build, Test, Deploy, Monitor lifecycle that turns experimentation into a repeatable practice.
Everything We Shipped at Interrupt
A full roundup of everything announced at Interrupt 2026 — from LangSmith Engine and SmithDB to Managed Deep Agents, Sandboxes GA, Context Hub, and LLM Gateway.
Introducing LangSmith Engine
LangSmith Engine watches your production traces, clusters failures into named issues, diagnoses root causes, and opens PRs with fixes — so your team reviews improvements instead of hunting for them.