Explainer: What to Know About Tech's Next Big Bet, AI Agents

AI agents, digital assistants pitted as the next evolution in artificial intelligence, promise to handle tasks autonomously, from booking reservations to negotiating business deals. At OpenAI’s DevDay event in San Francisco last week, the company demoed a voice bot calling a local delivery service and successfully ordering 400 chocolate-covered strawberries for the attendees, specifying the delivery and payment terms with barely a hitch. “2025 is when agents will work,” said OpenAI CEO Sam Altman. While this sounds futuristic, companies like Salesforce, HubPost, and Pactum AI are already deploying their own AI agents. But brand examples, like Qantas Airways, are still few and far between, a topic debated at Advertising Week New York. Here's a breakdown of these agents and the impact they will have. What exactly are AI agents? AI agents are more than just fancy chatbots. According to Parasvil Patel, a partner at Radical Ventures, AI agents don’t have one unifying definition and cover a broad range of functionalities, from automating workflows to scheduling meetings for you. But the goal is clear: “The ultimate aim is to execute work autonomously,” said Patel. AI agents today are in what’s called the co-pilot phase, handling specific tasks like summarizing meetings. But the real game-changer is when they transition to autopilot, managing more complex duties without human input. According to Patel, this leap could take up to 24 months. When did they emerge? AI agents first made waves on social media in early 2023 when several startups, such as AutoGPT—an open-source app built on OpenAI’s models—promised autonomous capabilities. “Many of those [experiments] were the brittle kind of agents that didn't scale and weren't able to put in production,” said Patel. How are companies using AI agents? The appeal is clear: AI agents can save time, boost efficiency, and relieve employees of repetitive tasks. A large distribution company grappling with managing 100,000 suppliers used Pactum’s AI, which deploys autonomous agents for negotiations. Instead of seeing negotiation as a dead end, these AI agents operated continuously to customize payment deals based on the speed at which suppliers’ goods moved. Vendors with faster-moving goods were offered more favorable payment terms. These tailored negotiations resulted in price discounts, rebates, and allowances, according to Pactum’s president, Martin Rand. Salesforce has also seen some results with its AI agents. Its pilot program, AgentForce, launched with five clients—including OpenTable and global publisher Wiley—and achieved a 40% increase in case resolution compared to its previous chatbot for Wiley. At the firm's Dreamforce event, Salesforce deployed AgentForce with Ask Astro which helped attendees plan their schedules by suggesting sessions and making reservations. Salesforce’s chief marketing officer, Ariel Kelman, told ADWEEK that the company invested heavily in developing its AI agent platform in response to client demand. "What companies are figuring out with generative AI is how do I get it to actually deliver productivity improvements for my employees and do useful things with customers," he said. What about roadblocks? The path to fully functional AI agents has challenges, particularly because managing different data formats—text, images, and videos—is complex, according to William Chen, director of product management for AI & emerging tech at Agora. “Your system is only as good as your data source,” Chen said. For Salesforce, the challenge is that customer adoption is still nascent, with companies just beginning to explore how to effectively utilize AI agents for productivity, according to Kelman. The key challenge, he said, lies in identifying what works best for employees and customers across various use cases. Are jobs on the line? Not exactly. AI agents won’t be taking jobs—not yet—but allowing employees to focus on more meaningful, strategic tasks, according to Rand. “The task of people will become to configure the autopilot, rather than flying the plane, and I think that’s a good thing,” he said. For example, a major logistics client of Pactum's previously relied on human negotiators to manage both short-term and long-term deals with freight providers and truckers. Now, AI agents can handle these negotiations more efficiently. The true economic value comes from the company’s ability to dynamically shift its business strategy “If the Suez Canal is backed up again and there’s a gloss of trucking services, they can move into long-term deals for better prices,” said Rand. “If we see a post-Covid shortage of logistics services, they can shift to short-term, spot rate deals to secure capacity.” What’s next? While companies are seeing early wins with AI agents, there’s more to explore. Salesforce plans to launch its next AI agent later this month: a Sales Development Representative (SDR) designed to manage early-stage sales interactions. Typically, human SDRs follow up on marketing leads through emails and calls, but this AI agent will qualify leads, presenting human salespeople with a refined list of 50 to 100 prospects who are eager to engage. “Instead of a list of 500 leads, they'll receive a targeted list of those who actually want to talk,” said Kelman.

AI News

Explainer: What Marketers Need to Know About Tech's Next Big Bet, AI Agents

Companies like Salesforce and OpenAI have been investing in AI agents

Trishla Ostwal

About

Subscriptions

Events

Publications

WORK SMARTER - LEARN, GROW AND BE INSPIRED.

Subscribe today!

Trishla Ostwal