The 4 Stages of AI Adoption—and Why Most SMBs Are Still Stuck at Level 1

It has been easy to focus that question on larger companies, given their scale and stakes. But it’s equally relevant for small-to-medium (SMB) businesses, from mom-and-pops to venture-backed startups. For this article’s purposes, we’re talking about companies with fewer than 5,000 employees; but we’re seeing the applicability of AI across the board—whether the business makes software or soup.

The capabilities of frontier AI models are accelerating at a pace that’s difficult to overstate. OpenAI’s GDPval benchmark finds that today’s best models are already approaching the quality of work produced by industry experts—completing professional tasks roughly one hundred times faster and at a fraction of the cost. And these gains aren’t incremental. Performance has more than tripled in just over a year. Meanwhile, younger professionals are already AI-native, cheerfully using these tools at work whether their bosses realize it or not. (They don’t realize it.)

It’s understandable many SMB leaders are reluctant to dive in. They’re not sure which tools or models to use and lack dedicated AI expertise. The landscape changes what seems weekly as a new model drops with a Marvel-villain-sounding name. But it’s far better to push through the overwhelm and start somewhere than deny or delay—because competitors won’t wait. That’s especially true in the current, uncertain environment, where every efficiency gain and competitive edge matters more.

Indeed, we’re seeing in many cases that AI-adopting SMBs are working to grow output and revenue without growing headcount (or replacing people). In other cases, AI-native founders are gaining 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 are and what might be holding them back.

The four stages of AI evolution

Successful integration of AI requires a structured approach. We’ve mapped four progressively more sophisticated stages to help leaders understand the value-creating roles AI can play. The goal is to locate where the business is today and consider how to move to the next level.

Level 1—Cog: The most basic implementation of AI, taking over previously manual work: rewriting emails, building customer lists, generating basic marketing copy. This is “fancy autocomplete.” It still requires human initiation and oversight. Most smaller businesses are here or at Level 2, if they’re anywhere on the map.

Level 2—Intern: Here, AI takes on more-sophisticated tasks: drafting initial proposals, triaging customer inquiries, generating first-pass budget forecasts. These are generally things the intern’s supervisor could handle, but AI creates efficiency and reduces labor costs. It’s an intern who never calls in sick or asks for a letter of recommendation. But it still needs you as the human to direct and guide it at each step, which can get tedious fast.

Level 3—Collaborator: Now AI operates as a true peer: analyzing cost structure, identifying pricing opportunities, building products, pressure-testing go-to-market strategy. It acts as a thought partner, surfacing insights an intern likely couldn’t. A good collaborator, like a good co-founder, gets better the more context it has. But it still requires frequent, in-depth interaction for best results.

Level 4—Agent/Service Provider: At this highest stage AI functions as a specialist or contractor, using tools in a loop to automate complex work that would normally require a dedicated individual/team: running end-to-end bookkeeping, managing customer-onboarding workflows, optimizing marketing campaigns across channels. At this level, AI is part of the business model, not just augmentation. It works largely independently, though humans remain in the loop. Getting here represents the largest evolutionary leap, and AI is increasingly up to the task: its capabilities are doubling roughly every seven months, compared with the two-year cycle of Moore’s Law for semiconductors. For context, that means the AI as SMB leader dismissed as “not ready” today may be twice as capable within half a year.

What’s stopping SMB leaders?

The stages, while easy to understand, are hard to implement, with several natural barriers at the starting line and on the way up.

Denial. Many leaders are still in denial about the need for AI, and that denial can take subtle forms, like assuming employees will figure it out on their own (they won’t; they’ll use it to write LinkedIn posts and call it a day.) Openness is the first step, but so is acknowledging that evolving use of AI isn’t easy.

Lack of structure. AI isn’t a plug-and-play application, especially at the later stages. Without intentional processes, clear data inputs, and defined workflows, leaders face the classic “garbage in, garbage out” problem. The companies that get the most from AI treat it as a system, not a toy or shortcut.

Mistrust and cultural resistance. Even if leadership is on board, the broader organization may not be. Making real progress requires trust and embracing opportunity that cascades company-wide. This is especially challenging because the progression isn’t linear—it’s exponential, with AI taking on increasingly complex tasks that require access to business context and data. Teams can overfocus on risk and get stuck on cautionary tales like Samsung’s well-publicized data leak when employees shared sensitive information on ChatGPT. Indeed, the U.S., collectively, is more concerned than excited about AI. But the larger risk, by far, is not using AI at all.

Inertia. Perhaps the most underestimated barrier 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 when running the business and putting out daily fires consumes all available bandwidth. But waiting for the perfect moment to start is itself a decision—and it’s usually the wrong one.

Ultimately, leaders are navigating two curves at once: progressing the role AI can play, and managing the emotional changes everyone must go through to be open to it. Technology is the easy part. People are always the hard part, due to the multiple types of friction we experience as barriers to any big change.

Start wherever you are

We understand if SMB leaders feel alarmed or unsure about AI. But we also hope they feel a sense of urgency and empowerment about what’s possible. The world has made a massive bet on AI: data-center investment was responsible for the vast majority of U.S. GDP growth in the first half of 2025, and the infrastructure buildout shows no signs of slowing. This is not a passing trend. It’s the new foundation of the economy.

The businesses that thrive in this next chapter won’t necessarily be the biggest or best-funded. They’ll be the ones whose leaders started somewhere, with an open mind, and kept going—from skepticism, to curiosity, to the realization that AI isn’t coming for their business. It’s coming for the businesses that pretend it doesn’t exist.

*

This article originally appeared in Inc.