Ride-hail giant Lyft has partnered with AI startup Anthropic to build an AI assistant that handles initial intake for customer service inquiries for both riders and drivers.
It’s the first phase of a broader collaboration between the two companies to use Anthropic’s services to research and test new Lyft products and build software internally.
The tie-up comes four months after Uber announced a similar partnership with OpenAI to launch an AI-powered assistant that can answer driver questions about EVs. More recently, Uber announced a collaboration with OpenAI’s new AI agent called Operator to help customers automate ordering food or scheduling rides.
Lyft quietly incorporated Claude, Anthropic’s family of large language models, into its customer care AI assistant in late 2024 via Amazon Bedrock, according to Anthropic. It provides answers to common support issues and redirects customers to a human specialist for more complicated tasks.
This isn’t Lyft’s first time with an AI chatbot; the company first introduced one to help triage customer complaints in 2018. But as anyone who has ever had to rely on a chatbot can tell you, old-school chatbots are limited and infuriatingly bot-like. Claude is known for its human-like responses, which could make all the difference to a customer seeking a resolution to a problem.
Or, it could be just another chatbot that allows companies to spend less money on customer service by forsaking the human assistants who could actually solve the problem much faster. Despite the promises of generative AI, most people still don’t want to talk to a bot when they have a problem to solve. A recent Gartner survey found that 64% of customers would rather companies don’t use AI for customer service. Another study found that when customers see they’re talking to an AI chatbot, it lowers emotional trust.
Still, Lyft says its Claude-powered assistant is working, and that it has reduced the average customer service resolution time by 87% and resolves thousands of customer requests every day. A Lyft spokesperson said that an issue is marked as resolved when a customer answers yes to the chatbot’s question of “Did we resolve your issue?”
Lyft’s nonexclusive partnership with Anthropic goes beyond chatbots. As part of the deal, Anthropic will provide Lyft’s engineers with training and education to teach them how to incorporate the AI company’s tools into their workflows.
Lyft uses AI to power its entire platform — everything from getting accurate ETAs and optimizing routes to choosing the right destination and coordinating rider pickup. Anthropic’s tools could help improve Lyft’s service so it can better compete with its main rival, Uber.
“Software engineering has undergone a seismic shift with the introduction of GenAI technologies. Gone are the days when humans were predominantly writing code,” Jason Vogrinec, executive vice president of platforms at Lyft, said in a statement. “With the promise of LLMs, especially leading models for coding like Claude, and agentic AI, we’re working to revolutionize our engineering organization to more effectively build game changing products for our customers.”
Anthropic also runs an “exclusive early access program” where certain customers can research and test new products. Neither Lyft nor Anthropic would share what such products might be, but an Anthropic spokesperson said Lyft’s feedback would help ensure that the company’s models and capabilities are “helpful for end users.”
Anthropic has raised $13.75 billion to date, per PitchBook data, including most recently $1 billion from Google. The company is in the process of raising another $2 billion at a $60 billion valuation.
This article has been updated with more information from Lyft about customer care resolutions.
Published by: Rebecca Bellan
