How AI Agents Will Disrupt Small And Mid-Sized Business In 2025

(This column originally appeared in Forbes)

Surveys are touting the “high adoption rates” of AI in small businesses. But this is just hype. Most of my clients aren't really using AI like their much larger counterparts. But that will change in 2025. Thanks to a wave of forthcoming AI agents, small businesses will finally be able to start automating with AI like big companies are doing.

While corporate brands like Klarna, T-Mobile, UBS Warburg and JP Morgan are spending millions to create large language models that can do everything from completing customer service inquiries, listening in on customer interactions, transacting investment purchases and behaving like human wealth managers, small businesses this year have been mostly tinkering around with generative AI chat bots like ChatGPT and Claude or toying with Microsoft Copilot and Google's Gemini for Workspace to help them write better emails or answer basic questions.

Here come AI agents

In 2024, the AI story was big businesses building big AI applications. That’s because smaller companies can’t afford to build. But in 2025 things will change. Instead of building their own applications, a wave of AI-enabled software will be available for smaller companies that will far surpass the capabilities of the chat bots we’re now using. They’re called AI agents.

An AI agent takes generative AI to the next level. It doesn't just generate a response or an email or a document. It actually does something. It performs tasks. It initiates transactions. It fixes problems. It behaves like a human.

Software companies large and small are investing in AI agents to enhance their offerings. Why? Because they understand that these are the types of features that will help their customers increase productivity and profitability. More importantly, they’re competing with each other to offer the kinds of tools that will keep their customers paying their monthly subscription fees. Good for them. And good for us.

Microsoft, for example, is rolling out ten agents for its Dynamics 365 customers. These "assistants" will better qualify leads by conversing with prospects, confirm sales and purchase orders, reconcile invoices in one ledger to cash receipts in other, approve expenses, resolve issues, close tickets and schedule field service agents. The company plans "many more agents" in the coming year.

CRM giant Salesforce is introducing SDRs (sales development representatives) and Einstein Coach agents that will also qualify leads before a human interacts, schedule meetings, and then offer a video likeness of a person resembling the lead to help the salesperson rehearse their pitch through role-playing and by providing competitiveness analysis before actually conversing with their actual human target. According to the company, the "SDR gets you meetings, the coach helps you get better at your pitch."

According to Intuit, which dominates the small business accounting software world, it's soon-to-be-released agents will be "bringing together the power of GenOS, the orchestrator, large language models, GenSRF (Security, Risk and Fraud), the UX experience, and it’s tying all that with what people really want, which is the ability to have work done for them.” Its financial agents will analyze a company's cash flow and then pay bills accordingly. Those same agents will digest and understand emails, documents and images to automatically process invoices and apply cash.

In the healthcare world, chipmaker Nvidia announced voice-enabled AI agents that will assume the role of nurses at a much lower rate per hour. In the software development world, firms like You.com have raised money to offer their "research" and "genius" agents that can retrieve reports and solve complex problems and better understanding complex scientific concepts.

For both consumers and small businesses both Anthropic and Google have announced new agents that can literally take over a device and then perform various tasks like web browsing, button clicking and text input to perform anything that an employee does (i.e. "book a flight on this day and this time" or "order more office supplies" or "submit a customer service complaint") on a single command.

Moving beyond generative AI to AI Agents

In 2025 small businesses won't just be using AI to "generate" a response to their queries. They'll be leaning on AI agents to actually do something with that information.

The big worry, as always, is people. Although none of these companies will publicly admit it, their agents are doing the tasks that humans are doing which obviously means that less humans will be needed. There is a great fear among employees that AI is going to replace them and that they’re going to lose their jobs. This is a legitimate fear.

For a business owner who's already facing labor shortages and struggling to increase their company's output and meet their customers' demands without increasing their overhead, agents will be useful. In an economy challenged by demographic declines and immigration hurdles, they're a Godsend. And until these agents are paired with technology like Google's Project Astra and robotics made by companies like Boston Dynamics and Agility Robotics, warehouse workers are still going to be in high demand. But that won’t take too long – probably a couple of years.

Should employees be concerned? Only the ones who don't embrace this stuff to make themselves more valuable. AI is like any other technology. It’s a productivity tool. But business owners have a responsibility to allay this fear by providing training and helping employees understand its benefits.

The reality is that agents are finally going to start moving AI from the corporate boardroom to Main Street this year. It won't happen immediately. They won't work as well as promised (when does a new technology ever?). But rest assured that, like the cloud, mobile apps and other technologies, our software providers will be pushing their customers to use their agents and my smartest clients will do just that.

What should small business owners be doing right now? That’s easy: talking to their software vendors about AI agents. While big companies are building their own internal AI driven applications, the tech industry is investing hundreds of millions building similar functionality for their customers. It’s up to us to understand these capabilities, learn them and apply them to our businesses in 2025.