Salesforce has its own. So does Microsoft. Amazon has recently sauntered into the ring, too, joining a cadre of other tech giants when it announced the release of its highly anticipated Nova Act.
I’m talking about AI agents, the biggest development to hit the tech world since Open AI released ChatGPT at the end of 2022.
More than just sophisticated chatbots, agents are changing how we think about AI and its capabilities. Leaders across industries are actively investigating how these autonomous, super-smart new systems can unlock unprecedented value, streamline operations and do everything from ease supply chain bottlenecks to design social media campaigns.
In fact, one study found that among IT and business executives, more than half are already using AI agents in their workflows, and another 35 percent plan to integrate agents by 2027.
But building an AI agent is not as simple as it may sound, and how leaders implement this new tech matters. Here’s what you should know.
Get Clear On Their Use
Let’s face it—there’s a lot of hype around AI, and in some cases, not a lot of clarity. In March, Tech Crunch published an article announcing that “No one knows what the hell an AI agent is.” The Wall Street Journal had a similar complaint, opting for the headline, “Everyone’s Talking About AI Agents. Barely Anyone Knows What They Are.”
I don’t completely agree. AI agents are generally understood to be a form of artificial intelligence that can simulate human reasoning. Unlike automated chatbots, agents can carry out complex conversations based on vast quantities of training data and other information sources. And unlike generative AI, agentic AI is proactive: It not only understands instructions, but can act autonomously to achieve goals.
As tempting as it is to jump on the agentic bandwagon, rushing is not the way to go. Worse still—don’t release a souped-up, AI-enabled chatbot and label it an “AI agent.” This sort of faulty marketing is precisely why the waters are currently so muddied.
Instead, take a hard-nosed, data-driven approach to really investigate how agents can be of use to your organization. Are you running a small business that needs to beef up its customer service, but lack the resources to hire around-the-clock staff? Do you head up a law firm and need administrative help with tedious tasks like pulling case details from intake forms and drafting filings?
These are exactly the types of jobs that agents can help with. But you wouldn’t hire someone to do a job without first understanding the role you want them to fill. The same goes with AI agents.
Build Trust First
If you’re following AI news, you know there’s a lot of concern about the ethical and safety implications of agents. Amid the valid issues raised about job displacement and over-reliance, there’s also wariness when it comes to trusting agents to act autonomously.
Kush Varshney, a fellow at IBM Research, points out that these potential ethical risks are very real. “Because AI agents can act without your supervision, there are a lot of additional trust issues,” he said. “There’s going to be an evolution in terms of the capabilities but also in unintended consequences. From a safety perspective, you don’t want to wait to work on it. You want to keep building up the safeguards as the technology is being developed.”
Because agents are privy to sensitive data and internal processes, leaders need to reassure employees, customers and stakeholders that agents are both secure and accountable—and for that to be true. If an AI agent is integrated into your sales funnel, for example, teams need to trust that it won’t share confidential pricing information with unauthorized users. If it’s handling customer inquiries, those customers need to trust that their personal data is being used responsibly.
Part of building that trust is through transparency, particularly around how AI agents are trained, what data they access and how decisions are made. Agents must also be tested and regularly audited to catch errors and biases in decision-making.
Doing this can not only head off compliance risks, but also bolster confidence in the technology. Think of it like putting new employees on a probationary period: you want to see how they perform under realistic conditions, give them feedback and then decide how to integrate them more broadly. In the same way, a pilot program for your AI agent—complete with clear metrics and oversight—can help build trust before a large-scale roll out.
Ultimately, trust is earned over time. A well-defined launch strategy and thorough, ongoing testing underscore your commitment to responsible AI.
Final Thoughts
AI agents are changing the way work is done, and their capabilities have expanded beyond what was possible just months ago. But as powerful as they are, leaders should not plan to deploy them overnight. A good rollout strategy involves being very clear on how they’ll be used; building a reservoir of trust among employees, customers and stakeholders; and ensuring that they are carefully monitored.
Read the full article here
