It’s never been easier to generate a social post, draft a newsletter, or spin up a polished website using AI like it is today. And many small businesses are embracing AI, automating marketing tasks, building faster sites and scaling campaigns that once required full teams.
But according to two executives shaping AI adoption for small businesses — Itai Sadan, cofounder and CEO of Duda, and Louis Gutierrez, director of AI at Constant Contact — a more complicated picture is emerging. While the excitement about AI-powered automation is real, most SMBs are not yet ready to combat the challenges that come with using AI.
Reality Versus Expectations
At Duda, which powers more than a million small business websites, nearly 45% of new sites now use at least one AI assistant, according to Sadan. From SEO optimization to content generation, the tools are being adopted at a striking rate.
“We’re seeing this huge uptake, especially in areas like image alt text generation or SEO automation,” Sadan said. “What used to take hours can now be done in minutes with a single click.”
Constant Contact is seeing a similar pattern. Gutierrez noted that many of their small business users are “doing more with less” — wearing multiple hats, streamlining content workflows and using AI to personalize customer communication at scale.
“AI lets them scale, yes,” Gutierrez said, “but it also allows them to focus on what they’re passionate about and delegate the rest.”
But these gains come with a shift in customer expectations. As AI makes marketing faster and more accessible, customers now expect more personalized experiences, faster turnaround and seamless execution. The pressure on small businesses to perform like enterprises — without the same resources — is growing by no small margins.
Real Infrastructure Needs
While AI appears magical on the surface, the infrastructure underneath is anything but. Gutierrez, who also teaches a course in explainable AI at UT Austin, likened AI models to perishable goods.
“Think of models like a can of beans,” he told me. “There’s an expiration date even if it’s not visible. Over time, their performance degrades. And without infrastructure that allows you to update, retrain, or replace them, you’re flying blind.”
That’s why Constant Contact’s biggest AI bet isn’t a shiny front-end feature but infrastructure. From fine-tuning pipelines to monitoring model drift, the company is building the capacity to stay agile regardless of whether today’s dominant model is replaced tomorrow.
Sadan sees infrastructure as equally essential. Duda recently rolled out MCP-powered AI copilots that allow users to launch, maintain and edit websites using natural language. In other words, users can now talk to the website builder like they would a human assistant — asking it to update content, fix layouts, or add features just by typing in plain language. But under the hood, it works by chaining API calls, model queries and platform logic into one coherent action layer.
“These are real agentic systems,” Sadan said. “But if you don’t have the data structures, APIs, and schema mapping in place, none of it works. The AI can’t fake its way through broken plumbing.”
Governance Isn’t Optional Anymore
Despite the impressive front-end capabilities, both leaders warn that AI without governance is a disaster waiting to happen, especially for small teams with limited oversight.
“At some point, a staff member will drop sensitive customer data into ChatGPT,” Sadan said. “Not because they’re careless but because there are no rules.”
Gutierrez agreed, noting that that’s why Constant Contact established an internal AI governance committee to vet AI tools for ethics, privacy, security and standardization. The challenge, he explained, isn’t just performance. It’s what happens when tools fail or deliver biased results.
“Everything’s easy when you prototype on your laptop,” he said. “But the second a model goes into production, it behaves in ways you never imagined. That’s where governance matters most.”
Without a clear policy, companies risk leaking data, overpromising capabilities, or baking systemic bias into customer-facing tools. And while governance may sound like an enterprise luxury, both executives say it’s a small-business necessity.
The New SEO War
Perhaps the most overlooked trend in small business AI is how people find products and services in the first place. While Google still dominates traffic, Sadan said traffic from ChatGPT citations is “growing at a staggering pace.”
Duda is betting heavily on this shift. In addition to automating SEO tasks, the company is investing in AI discoverability features that optimize websites for how LLMs index and interpret them. From FAQ schemas to product data markup, the goal is to rank in the age of AI-powered search, not just traditional queries.
“This is like the early days of Facebook for businesses,” Sadan said. “If you understand how to ride the wave early, you can grow faster than your competitors, all for less.”
Gutierrez, meanwhile, is watching another shift: Vibe coding — the ability to build full software applications using natural language. “It’s turning non-coders into full-stack developers,” he said. “In a few years, every small business could build their own CRM or booking system just by describing what they want.”
Wielding AI Well
Both executives agree: AI won’t replace small businesses, but using it poorly could put them at risk. Gutierrez warned against the current trend of overpromising what AI can do. “I’m in vendor meetings all the time,” he said. “Too many pitch AI as a magic box that does everything. That’s dangerous.”
Sadan echoed this sentiment, adding that the biggest mistake small businesses make is adopting tools without evaluating their actual productivity gains. “Try the tool,” he said. “If it doesn’t make your day faster or your work better, it’s not worth it, no matter what the marketing says.”
What’s missing from many AI adoption stories is restraint. It’s not enough to use AI; you have to understand what it’s good at, what it’s not and how to build around that reality. That means investing in infrastructure, enforcing governance and designing for a future where autonomous discovery and interaction may define who thrives and who doesn’t.
As Gutierrez put it: “You don’t win by adding AI to your toolset. You win by building a business that knows how to wield it.”