AI agents have rapidly advanced in the enterprise and the amount of work they do will only continue to grow.
The main questions that remain unanswered are how humans will adapt to their AI agent colleagues, and whether an agent's output can be trusted.
These questions and the rapidly evolving ecosystem around agentic AI were prominent at the AI Agent Conference 2026 in New York City from May 4-5.
There's little doubt that AI agents are transforming the enterprise, said Arvind Jain, founder and CEO at Glean in a conference session. The biggest disruption is in labor-intensive areas, such as customer experience or in more deterministic areas, like coding.
However, the AI agents of today are not like software systems of the past, he said.
"The difference between software subscriptions of the past and agents is that agents are about doing the tasks," Jain said. "We're talking about coworking with an agent working as a person."
There will be a lot of work for AI agents in financial analysis, as they are becoming increasingly effective at the tasks financial analysts perform, said Jai Das, president and partner at Sapphire Ventures, during the session.
"Slowly, the models and agents will get better at doing some tasks, and then enterprises will have to learn how to take these models and build on top of them," Das said. "We are just in the early innings of enterprises understanding how they can build harnesses to create these agents to do the work based on the models."
However, enterprises will soon have to consider the rising cost of these models, he said.
"It's going to be a big issue to use all the tokens you'll need for development," Das said.
Companies will need to transform into AI-native organizations to survive in the agentic world, he said. The ones that make a successful transformation to being AI-native will need to change their business model, their code development and their go-to-market. And the transition must be driven by a strong management team and CEO.
"They have to be fully behind the effort to make the company AI-native -- not only from a business perspective, but also how it's developed, sold and managed within the company," Das said. "The companies that will make it are the ones that take a radical approach rather than small steps."
T-Mobile goes AI-native
Telecommunications giant T-Mobile is transforming into an AI-native enterprise. Julianne Roberson, director of AI engineering at T-Mobile, said in a conference session that T-Mobile has gone all-in on AI since 2024.
T-Mobile worked with OpenAI and AI vendor Distyl AI to develop an application called IntentCX that used AI to help the company understand customers' issues and intentions to solve problems before they happen, Roberson said.
"We get a ton of calls through customer care and wanted to take those out of the business, not by forcing people to not talk to a human but by driving a high NPS experience and solving the customer's problem before they face it," she said.
Customers leave breadcrumbs of why they leave T-Mobile, and IntentCX's AI is tasked with helping to better understand these breadcrumbs so they will stay with T-Mobile longer if solved, Roberson said.
"We put bots on the app and website, have voice bots in the app and voice AI answering the phone for more than half of phone calls now," she said.
T-Mobile's customer-centric culture is important for the success of the AI-native transformation, and Roberson said it's critical that company leadership is all-in on AI.
"Any time there's a roadblock that we face, the leaders are in a room with us. We have one-on-one texts with the CIO and meetings with the CEO," she said. "It's all about how to move forward, and they understand that it's not easy. You can't just put AI out there."
AI coding agents take off
Agentic AI is transforming the world of coding, and this could have implications for enterprise SaaS software vendors, said Raghu Malpani, chief technology and product officer at automation vendor UiPath, in a conference session.
Malpani said that as coding agents take over from SaaS vendors, more processes can be automated and managed by agents.
"We're going all in on coding agents. We want to make our platform available as APIs and CPIs where you can make the best out of it by automating the building and managing all of the software development lifecycle," he said. "The obvious question is, if a coding agent can build software from the ground up, why would we use a SaaS vendor?"
One reason is governance, because SaaS vendors provide auditability, traceability, compliance and certifications, Malpani said. However, the less obvious answer is that coding agents on SaaS platforms act more as composers than coders, which reduces development time.
"For example, UiPath has these building blocks, and the coding agent configures the platform to accomplish the process outcome as opposed to coding everything from scratch," he said.
Jim O'Donnell is a news director for TechTarget, where he covers IT strategy and enterprise ESG.
