Article in The New Stack by Eric Newcomer
With careful governance and validation, AI agents are increasingly gaining adoption for specific enterprise functions, according to speakers at the AI Agent Conference this week in New York.
The popularity of new, highly capable AI coding agents has grown dramatically over the past year. Still, the code they generate cannot be trusted in production, said Datadog’s Chief Scientist, Ameet Talwalkar, in the opening keynote.
“One of the hardest things for humans to do is no longer building production systems. It’s actually reviewing the vibe-coded software that gets shipped into production,” Talwalkar said.
Datadog is extending its observability product line to model real-world systems and predict production issues with AI agents before they happen, he said.
The most popular use of AI agents in business applications is for customer service and customer assistance chatbots.
T-Mobile, for example, uses AI agents to handle 200 thousand customer conversations a day, said Julianne Roberson, Director of AI Engineering at T-Mobile. This project took about a year to complete, she said.
Zhou Yu, co-founder and CEO of ArklexAI, an agentic framework supplier, told The New Stack that the company’s new ArkSim product aims to shorten time-to-market for customer-facing bots by simulating AI-agent interactions with customers. AtkSim collects data to improve quality, since agentic interactions are not deterministic.
