UpTrajectory Review

Fast Company's Darren Person delivers a necessary corrective to the AI hype cycle: the problem is not that small businesses are failing at AI implementation, but that they are succeeding at the wrong thing. The piece argues that most organizations treat AI as a faster horse—speeding up email summaries, automating routine tasks—rather than asking whether the carriage itself belongs in the modern economy. This is not a new insight in the abstract; business thinkers have been warning against 'paving the cow path' since at least the 1990s reengineering movement. What gives Person's version teeth is the specificity of the failure rate: 60 percent of companies report little value from AI initiatives despite significant investment. That is not a technology problem. It is a design problem, and small businesses are particularly vulnerable to it because they lack the organizational slack to absorb expensive mistakes.

For the small-business operator, this framing matters intensely because the resource being wasted is not just money—it is attention and organizational capacity. A five-person shop that spends six months layering AI onto an invoicing workflow has burned something irreplaceable: the founder's focus on customer relationships, product development, or market positioning. Person's example of his own team's prototyping capacity is instructive here. The shift from two or three concurrent ideas to dramatically more is not merely quantitative; it changes what the team considers possible. But notice the precondition: the team did not shrink. Small businesses must resist the immediate temptation to treat AI gains as labor substitution, because that path leads to the same processes with fewer people, not better outcomes with the same or more people. The competitive advantage lies in output quality and speed, not headcount reduction.

What is genuinely contested in this argument, and where we are partially skeptical, is the implied ease of the 'rethink from the ground up' prescription. Person acknowledges that people closest to the work are trapped inside existing processes, but his solution—pairing them with teams who 'understand that AI creates an opportunity to challenge assumptions'—assumes such teams exist and are accessible. For most small businesses, they do not. The consultant or internal strategist who can credibly ask 'would we design this workflow today?' is a luxury good. The piece also elides a harder tension: sometimes incremental efficiency is the right call because the business model depends on it. A local accounting practice that automates client onboarding may not need to rethink accounting itself; it needs to free partner hours for higher-margin advisory work. Not every workflow merits existential interrogation.

The downstream effects of Person's framework, if adopted selectively, reshape competitive dynamics in ways small businesses should monitor. First, the gap between 'AI-augmented' and 'AI-native' operations will widen faster than expected. A company built from inception with AI-integrated prototyping, customer research, and product development will move at a different clock speed than one retrofitting tools into legacy structures. Second, the labor market for hybrid roles—part domain expert, part AI workflow designer—will tighten. Small businesses that cannot offer the scale or compensation of larger competitors may need to invest unusually heavily in training existing staff rather than hiring from outside. Third, and less discussed, the vendor ecosystem will respond to this demand with 'transformation' consulting that is merely repackaged implementation. Discernment in purchasing will become a core operational capability.

What to watch: whether the 60 percent failure rate Person cites becomes a catalyst for vendor accountability or merely a talking point. The AI tool market is currently optimized for easy deployment, not structural redesign. Small businesses should expect their vendors to demonstrate not just speed gains but outcome changes—and should demand proof. What to do now: audit one workflow where you have already deployed AI and ask Person's question explicitly. If you were building it today, would the steps be the same? If the answer is no, the sunk cost of the current tool is less important than the compounded cost of continuing. The harder discipline is stopping projects that are merely efficient, not transformative. That requires judgment this piece rightly treats as scarce and worth cultivating.

Person's argument ultimately lands where serious operational thinking always does: technology is not strategy, and efficiency is not impact. The small-business reader should take from this not a mandate to rip and replace every system, but a permission structure to be dissatisfied with incrementalism. The businesses that will matter in five years are those that used AI to ask harder questions, not those that asked the easy ones faster.

Takeaway: Audit one AI-enabled workflow and ask: if building it today from scratch, would the steps be identical? If not, redesign rather than optimize.