It was easy to focus this question on larger companies given their scale and stakes. But it’s just as important for small and medium-sized (SMB) businesses, from mom-and-pop to venture-backedstart-ups. For the purposes of this article, we’re talking about companies with fewer than 5,000 employees. but we see its applicationAll includedgenerally—whether the business makes software or soup.
The capabilities of frontier AI models are accelerating at a rate that is hard to overestimate.OpenAI’s GDPval benchmarkfinds that today’s best models are already approaching the quality of work produced by industry experts—completing professional tasks about a hundred times faster and at a fraction of the cost. And these gains are not incremental. Performance has more than tripled in just over a year. Meanwhile, younger professionals are already native to AI, happily using these tools at work, whether their bosses realize it or not. (They don’t realize it.)
Understandably, many SMB leaders are hesitant to take the plunge. They are not sure which tools or models to use and do not have specific expertise. The landscape changes what seems like a weekly basis as a new model drops with a name that sounds like a Marvel villain. But it’s much better to get over the hump and start somewhere than to decline or delay—because competitors won’t wait. This is especially true in the current, uncertain environment, where every efficiency gain and competitive advantage matters more.
Indeed, we see in many cases that SMEs adopting AI are working to increase production and revenue without increasing headcount (or replacing people). In other cases, native AI founders gain traction with enterprise customers using Claude code, a full stack of agents, and just two to three employees!
This article maps 4 stages of AI evolution to help business leaders understand where they stand and what might be holding them back.
The four stages of the evolution of artificial intelligence
Successful AI integration requires a structured approach. We’ve mapped four progressively more sophisticated stages to help leaders understand AI’s value-creating roles. The goal is to identify where the business is today and think about how to take it to the next level.
Level 1 — Cog: The most basic application of artificial intelligence, taking over the previous manual work: rewriting emails, creating customer lists, creating basicmarketingcopy. This is “fancy autocomplete”. It still requires human initiation and supervision. Most of the smaller businesses are here or on Level 2 if they are anywhere on the map.
Level 2 — Practice:Here, artificial intelligence takes on more complex tasks: writing initial proposals, evaluating customer questions, creating budget forecasts for the first time. These are generally things that the intern’s supervisor could handle, but AI creates efficiencies and lowers labor costs. He is a boarder who never calls in sick or asks for a letter of recommendation. But it still needs you as a human to direct and guide it every step of the way, which can get tiresome quickly.
Level 3—Associate:Now AI acts as a true peer: analyzing cost structure, identifying pricing opportunities, building products, stress-testing go-to-market strategy. It acts as a thought partner, bringing out ideas that an intern probably wouldn’t be able to. A good partner, like a good co-founder, gets better the more context they have. But it still requires frequent, in-depth interaction for best results.
Level 4 — Agent/Service Provider: At this higher stage AI acts as an expert or contractor, using tools in a loop to automate complex tasks that would normally require a dedicated person/team: running ledgers, managing customer onboarding workflows, optimizing omni-channel marketing campaigns. At this level, AI is part of the business model, not just augmentation. It works largely independently, although people remain in the loop. Getting here represents the biggest evolutionary leap, and artificial intelligence is increasingly up to the task: its capabilities aredoubling approximately every seven monthscompared to the two-year cycle of Moore’s law for semiconductors. For context, this means that AI as an SMB leader dismissed as “not ready” today could be twice as capable in half a year.
What’s holding SMB leaders back?
The stages, while easy to understand, are difficult to implement, with several physical obstacles at the starting point and on the way up.
Refusal. Many leaders still deny the need for AI, and this denial can take subtle forms, such as assuming that employees will figure it out on their own (they won’t, they’ll use it to write LinkedIn posts and call it a day).
Lack of structure.AI is not a plug-and-play application, especially at later stages. Without purposeful processes, clear data inputs, and defined workflows, leaders face the classic “garbage in, garbage out” problem. Companies that make the most of AI treat it as a system, not a game or a shortcut.
Mistrust and cultural resistance.Even if leadership is there, the wider organization may not be. Achieving real progress requires trust and acceptance of opportunities that permeate throughout the company. This is particularly difficult because evolution is not linear — it is exponential, with AI taking on increasingly complex tasks that require access to business context and data. Teams can focus too much on risk and get stuck on cautionary tales likeThe much publicized Samsung data leakwhen employees shared sensitive information on ChatGPT. Indeed, the US, collectively, ismore worried than excited about artificial intelligence. But the biggest risk, by far, is using no AI at all.
Inaction.Perhaps the most underrated obstacle is simple organizational inertia—the gravitational pull of “how we’ve always done things.” Even leaders who see the opportunity can struggle to prioritize AI while running the business, and putting out daily fires eats up all available bandwidth. But waiting for the perfect moment to start is a decision in itself—and usually a wrong one.
Ultimately, leaders are navigating two curves at once: advancing the role AI can play and managing the emotional changes everyone must undergo to be open to it. The technology is the easy part. People are always the hard part because of itmultiple types of frictionwe experience as obstacles to any great change.
Start wherever you are
We understand if SMB leaders feel anxious or uncertain about AI. But we also hope they feel a sense of urgency and empowerment about what is possible. The world has made a huge bet on artificial intelligence:Data center investment was responsible for the vast majority of US GDP growth in the first half of 2025and infrastructure development shows no signs of slowing down. This is not a passing trend. It is the new foundation of the economy.
The businesses that thrive in this next chapter won’t necessarily be the biggest or the best funded. They will be the ones whose leaders started somewhere, with an open mind, and continued – from skepticism, curiosity, the realization that AI is not coming for their business. It’s coming for businesses that pretend it doesn’t exist.
*
This article originally appeared on Inc.
