Real AI Transformation Requires Redesigning The System, Not Replacing the Power Source

When generative AI first hit the enterprise, the corporate math looked simple: if an algorithm can execute a task, the person doing that job is redundant. Top-down pressure from boards and investors to show immediate returns from emerging technology resulted in rushed workforce reductions. But as companies cut jobs based on AI's potential rather than its proven performance, many organizations hit a wall and accidentally broke their own operations. The result is a growing pattern of companies reversing course, rehiring roles they eliminated, and discovering that the knowledge of how the work actually got done left the building with the people they let go.

Watching the pattern unfold is Jen Spencer, Chief Growth Officer at the global talent and managed outsourcing company Booth. Before joining Booth she was CEO of SmartBug Media, which she scaled from 26 to more than 300 employees through multiple acquisitions and organic growth driven by a strong demand gen strategy. Her vantage point, sitting between companies making these cuts and the talent decisions that follow, gives her a direct view of how AI-driven restructuring is actually playing out.

"Saying, 'We're going to have AI do it,' left many companies in a worse position than when they started, because they didn't have intimate knowledge of how the work actually gets done. That knowledge was living with the people that they decided they didn't need anymore," she points out. The gap between AI's promise and its reality is where the strategy problem lives, and solving it starts with examining how the decisions were made in the first place.

The slippery slope of "if this, then that"

Spencer traces the pattern to the moment generative AI went mainstream and leaders began making fast, untested logical leaps about what it could replace. "As soon as generative AI came out, they said, 'AI can write a blog. Then why do we need writers anymore?' It started this slippery slope of, 'If this, then that,' without really being logic-checked, gut-checked, or reality-checked."

The disconnect became starkest at the industry events Spencer attended to track how businesses were approaching transformation. The people building the technology and the people under pressure to deploy it were operating on completely different timelines. "All these AI vendors were saying, 'Don't expect ROI from AI in the first year. You have to understand your processes and your systems, and most people don't have clean systems or documented processes, so applying AI right now is just going to make a bigger mess,'" she shares.

At the same event, the enterprise innovation leaders in the room were under a directive that contradicted exactly that guidance. "Every person there had a directive to see ROI from AI this calendar year, and the ROI had to come from actual cost savings, dollars saved, being able to do this thing you used to do without the people that used to be required." The vendors closest to the technology were counseling patience and groundwork. The boards were demanding immediate headcount savings. Leaders were caught in between, and the cuts happened anyway.

How much room a company has to get AI right depends largely on who owns it and what those owners are counting. A venture-backed business is usually measured on growth and customer acquisition, with investors willing to fund losses in pursuit of market share. A private-equity-owned company, on the other hand, is measured on efficient growth, where the scoreboard is capital discipline: customer acquisition cost that pays back inside roughly 12 to 18 months, net revenue retention above 120 percent, and pipeline coverage of three to four times the target. That tighter scoreboard is what turned AI from an experiment into a mandate.

Across B2B SaaS, the median company now spends about two dollars to bring in a single dollar of new recurring revenue, and the Rule of 40 has become the number investors watch. It's the idea that a company's growth rate and profit margin should together clear 40 percent. Fewer than a third of software companies actually clear it in a given year, and those that do have commanded valuations more than double their peers'. To a board holding that yardstick, a technology pitched as a way to strip out cost stops looking like a bet and starts looking like an obligation.

The missing map of how work gets done

The deeper problem, in Spencer's analysis, is that many organizations were not well balanced to begin with, and the AI directive gave them a reason to cut without first understanding what they were cutting. She frames the work inside any function as a spectrum of who or what should be doing it.

Some work can be fully automated. Some genuinely requires a full-time employee who lives and breathes the brand. Some is best handled by fractional strategic consultants or specialized agencies. And some is lower-risk work well-suited to global talent markets. The failure was collapsing all of that nuance into a single sweeping decision. "Everyone overcorrected on AI. They decided what people to keep, who to cut, and then whoever they kept, they'd have them do all the things they decided AI wasn't going to do. But they didn't take the time to think about what work needs to get done and what's the best way for it to happen," Spencer says.

Companies that had never used fractional or outsourced resources may have been carrying excess full-time headcount, but replacing that headcount with AI rather than rethinking the work left them worse off. The institutional knowledge of how the work actually happened was the asset they discarded without realizing its value.

Few processes make that point as sharply as the handoff between an SDR and an AE. Forrester found that companies with a formal handoff stage close 9.3 deals for every thousand inquiries, compared with 4.6 for those without one. On paper, the handoff reads like an administrative step: the SDR books a qualified meeting and the AE takes it from there. In practice, it's one of the most knowledge-dependent moments in any revenue organization, and almost none of what makes it work is written down anywhere. The unspoken signals that told the rep an account was worth chasing live in the people running the motion, not in a CRM field.

Swapping the power without redesigning the system

Spencer's sharpest illustration comes from an analogy about the shift from steam to electricity in early factories. It captures why swapping AI into an old operating model produces so little real transformation. "When electricity was first invented, factories didn't move the machines. They just replaced steam power with electric power. Innovation didn't happen until someone realized, 'Wait, this machine doesn't have to be here. The reason it's in this location has to do with something that doesn't even matter anymore.'"

The steam-era machines were positioned around a central steam engine's driveshaft and belt system that electricity made irrelevant with unit drive, but manufacturers kept the old layout because they were thinking in terms of substitution rather than redesign. The parallel to AI is direct. Dropping AI into the same workflows, the same team structures, and the same decision pathways changes the power source without changing the system.

In Spencer's view, real transformation requires rethinking where work happens, who should do it, and which bottlenecks no longer need to exist. The obstacle is that teams rarely have time for that ground-up rethinking while they're still running the business, and having the rug pulled out through premature cuts makes it even harder.

AI-native is becoming a hollow label

That substitution trap shows up clearly in revenue operations, where the language of transformation has outpaced the reality. Teams describe themselves as AI-native while still routing decisions through the same pathways that existed before, which means faster activity but the same decision latency. Spencer is skeptical of the label itself. "People say they're AI-native, but they're not. It's already an overused, misused buzzword."

Her alternative is to start from the outcome rather than the tool. Booth's approach, she says, never begins with AI. Instead, it starts with the end in mind and reverse-engineers the process to get there. The reason it matters is that the purpose of RevOps has not changed just because AI entered the picture. The goal is still to create a better customer experience, to move trustworthy information to the people who need it, and to reduce friction across go-to-market, operations, and finance. AI can serve those goals, but only for teams that first understand the work well enough to redesign it rather than cutting the people who understood it and hoping the technology fills the gap.