LaunchDarkly launches runtime control layer for the agentic AI era
LaunchDarkly, a feature control platform that helps developers and software engineers launch and manage products, today announced the launch of AgentControl, a new solution providing real-time management of AI agents in production.
As more agents get launched into production, companies are discovering that, unlike ordinary code, generative AI is probabilistic and doesn’t always behave exactly as it did in testing.
Although models can update and go through experiments, they might run into unexpected user interactions that change output quality without a single change to code. Once deployed, agents are given broad autonomous leeway, with less human oversight, primarily so they can get their job done. But that leaves them open to straying from their tasks as anomalies build up, and it can happen faster than teams can react.
“The hardest problems in AI, like model drift, unpredictable outputs, and the inability to intervene fast enough, turn out to be exactly the problems our platform was built to solve,” Chief Technology Officer Cameron Etezadi said. “We didn’t have to reinvent the platform; we just had to extend it to meet the demands of an AI SDLC and agentic-driven workflows.”
