Good morning. AI agents are still in the early stages and while enthusiasm over their potential remains high, for most companies it is way too early to capture a return on investment.

Beyond a few areas such as coding and customer service it is hard to find examples of them deployed at scale and driving returns.

That was the gist of many conversations I had – even among the self-selecting group that convened for the AI Agent Conference, hosted by FirsthandVC, in New York this week.

“Whenever you get a very specific question like, ‘can you give me an example of an agent that's deployed at scale that's actually moving the bar on your P&L,’ it's very hard to get an actual answer,” said Akamai CTO Robert Blumofe.

The key to realizing their value is often bound up in the corporate mindset and the way companies deploy agents, according to attendees I talked with. As companies update their processes with AI in mind, they will be in a better position to capture meaningful value from agents.

They should start with thinking about applying AI agents across the entire enterprise.

Datarobot Chief Product Officer Venky Veeraraghavan said companies should focus on end-to-end workflows, rather than automating particular jobs. Anthropic’s Claude Code was so successful, in part, because it integrated product developers, product managers and product designers into a single workflow. It didn’t just provide individual AI copilots to all the people in those roles to make them more efficient – which is the trap many companies are falling into now.

The problem isn’t with the technology; it’s that people don’t know what to do with AI agents which can be more complicated and less intuitive than a chatbot. The answer is that the AI needs to explain itself to the user, according to Glean CEO Arvind Jain. “People don’t know what AI can do for them, so they don’t try.” Glean, which provides a platform for AI at work, tries to get around that problem by prompting users, for example, offering to write the response to an email that’s sitting in their inbox.

Integrating AI effectively starts with putting deployment in the hands of the end user and not a centralized IT team, according to Peter Day, general partner at San Francisco-based venture studio Super{set}. “You need to empower the people closest to the problem to solve it,” he said. “If I was CEO of a business, and I wanted to agentify my finance processes, I wouldn't ask IT.”

What do you think? Do you agree with these reasons for the AI value gap or are you seeing something else? Email me at isabelle.bousquette@wsj.com and let me know!

Companies around the world are starting to think about these organizational changes related to deploying AI agents. Here is how ride-hailing company Lyft is approaching it.

Jason Vogrinec, Lyft’s executive vice president of AI transformation, sat down with the WSJ Leadership Institute to discuss where the company is putting AI agents to work—highlighting its work in customer service and software development.

More than half of Lyft’s employees are “builders,” according to Vogrinec. That includes software developers, product managers or data scientists, and for those employees, “AI is transforming the way we think about building software,” he said.

Even as a relatively young company—one that was founded following the advent of cloud-computing—Lyft needs to change the way it develops its platform to become truly “AI-native,” Vogrinec said.

When the company first began experimenting with AI in coding, human developers were being assisted with AI tools. Now, Lyft is shifting toward a model where humans delegate tasks to agents—essentially becoming “software development managers” rather than “software developers,” Vogrinec said. Tasks outsourced to AI agents include architectural reviews, product spec updates and code repository clean-up.

In customer support, AI agents are tackling about half of all the 15 million contacts Lyft handles each year. Those AI agents can resolve customer issues in under three minutes, Vogrinec said. The company has also developed specialized AI agents that handle issues like damage to a vehicle or lost items.

“We’re finding agentic development and agentic capabilities are here,” Vogrinec said, “And it’s now about finding the opportunities to leverage those at relatively small scale and seeing them expand over time.”

— Belle Lin

• CoreWeave posted larger Q1 losses and said it was raising its capex guidance this year to a range of $31 billion to $35 billion, up from a previous low end of $30 billion, citing higher component costs. 
• France is formally charging Elon Musk in connection with X's handling of illegal content, the latest example of a new strategy from cybercrime investigators in the Paris prosecutor’s office to hold leaders of tech companies criminally responsible for what happens on their apps.
• Anthropic is pursuing a funding round that would value it at nearly $1 trillion, the FT reports, positioning the AI startup to overtake OpenAI as the world's most valuable company in the sector.

The WSJ Tech Council brings together CIOs, CTOs and CISOs advancing innovation and shaping the future. Join this trusted community where tech executives connect with peers to explore emerging trends and gain the perspective they need to stay ahead of disruption.

Request Information

Follow Isabelle Bousquette on LinkedIn, Instagram, X, and TikTok for more behind the scenes on her tech and AI coverage, and lately, her contributions to the WSJ Leadership Institute's new Executive Resilience series, where she's profiling America's top execs about their fitness and wellness habits.

Follow Belle Lin on LinkedIn and X for her latest reporting on enterprise technology and AI.

Steven Rosenbush is chief of the enterprise technology bureau at the WSJ Leadership Institute. He also has a column. You can follow him on LinkedIn.

Tom Loftus is the editor of The Morning Download. He suggests following Isabelle, Belle and Steve on their various social channels. But if you insist, here's his LinkedIn.