What Steam Engines Teach Revenue Teams About AI: Capacity Lift, Not Headcount Cut

Gartner surveyed 350 global executives at companies above 1 billion in revenue already deploying autonomous AI capabilities. Roughly 80 percent had reduced workforce, and the ROI gap between those who cut the most and those who cut the least was nearly zero.

Gartner's lead researcher Helen Poitevin put it directly: workforce reductions may create budget room, but they do not create return.

That is the trap, and a lot of leadership teams are walking into it. The companies actually moving the number figured out which work to stop giving to humans, then pointed those humans at growth that was never reachable before. That distinction is everything, and it maps almost exactly to the invention of the steam engine.

When James Watt's steam engine arrived, engineers described its output in horsepower because it was the only unit people could conceptualize. Not because horses were the right analogy, but because humans needed a bridge to understand something with no precedent. The steam engine did not replace horses on the farm. It opened industrial capacity horses could never have reached at any scale or cost. Entire industries came into existence that simply were not possible in a horse-powered world.

AI in GTM is the same shift. The opportunity is reaching customers and moments that were never economical to serve with humans.

Replacing humans is the wrong bet, and the data proves it

The instinct to cut is understandable. AI promises efficiency, boards want returns fast, and layoffs are a visible signal that the company is taking AI seriously. Challenger, Gray and Christmas tracked 49,135 AI-attributed layoffs through April 2026 alone, and 56 percent of all 2026 layoff events now explicitly cite AI as the reason.

The returns are not there. Forrester's Predictions 2026 found 55 percent of companies now regret their AI-driven layoffs. Gartner predicts that by 2027, half of companies that attributed headcount reductions to AI will rehire for similar functions, often under new titles. Robert Half data shows 29 percent of companies that cut staff for AI have already reopened those exact positions.

Analysts call this the AI boomerang. Klarna cut 700 customer support roles for AI chatbots and reversed course within six months because customers did not like the experience. IBM replaced HR staff with AI and brought people back after similar quality failures. The financial trap compounds: enterprise AI platform bills scaling past a million a month wipe out salary savings, while rehired AI-native roles command 20 to 35 percent higher salaries than the positions they replaced.

Underneath all of it is a buyer experience problem. When companies cut the wrong people and hand judgment-intensive work to AI before it is ready, buyers notice. Response quality drops. Account knowledge disappears. Relationship continuity breaks. The buyer's experience of your company degrades at the exact moment AI investment is supposed to be improving it.

Gartner's data points at the framing itself. The companies generating the highest gains used AI as people amplification, equipping humans to expand what they could do, rather than removing them from the motion.

The honest tension between two of the clearest thinkers

Amanda Kahlow does not soften the forecast. In a 2026 interview she said there will be many jobs replaced across go-to-market, potentially 95 percent of them in the next two to five years. Her framing elsewhere: certain roles are gone, the SDR as we know it today is gone, the solutions engineer joining call number 17 because the buyer finally earned the right to real answers is gone. Not predictions. Happening now.

Her internal operating philosophy runs accordingly. She actively encourages her own team to find ways to replace their own roles with AI, with the promise that those who succeed get new, higher-value positions. In her words, there will be more jobs as a result of AI, we just need to reskill, and they will not be the same jobs. The work changes. The people stay and move up.

Jacco van der Kooij runs the same argument from a different angle. He has written that AI will automate roughly 70 percent of sales activities, specifically CRM entry, scheduling, call prep, and qualification at scale, while the remaining 30 percent requiring relationship expertise and judgment becomes more valuable by comparison. His firm proved it on their own revenue motion, deploying an AI SDR built over 12 months and running it in production: 2,030 qualified conversations, 100 percent CRM capture, and a 200,000 dollar deal closed.

He also gave GTM a framework worth keeping: the HITL curve, human-in-the-loop, borrowed from automotive engineering. The principle is that as the potential impact of failure increases, so does the need for human oversight. Low-risk tasks like call briefings, summaries, and initial qualification can run autonomously. High-stakes work like diagnosing complex problems, designing multi-stakeholder solutions, and navigating executive relationships requires human expertise with AI assisting. The HITL curve lets you make the handoff decision by consequence rather than by job title.

Where the steam engine analogy earns its weight

The steam engine did not put horse groomers out of work and leave everyone worse off. It created locomotive engineers, railway operators, conductors, factory workers, and entire industries of transportation and manufacturing that did not exist before. The net number of jobs went up. The type of work changed entirely. The people who understood the shift early retrained into the new work. The people who waited got left behind by timing.

MIT's NANDA Initiative found 95 percent of AI pilots fail to deliver P&L impact. Gartner found replacement strategies generate no ROI advantage. Both trace failure to the same root: the deployment was designed to shrink the team rather than reach customers the team could never have served.

The human optimization audit

Step 1: Map where human time is actually going

For each revenue-critical role, list the top ten activities that consume a week. Be granular. Not selling. Specifically: inbound triage, follow-up emails, call prep research, CRM data entry, scheduling, technical Q and A, deck customization, internal status reporting.

Most teams find that 40 to 60 percent of time in quota-carrying roles goes to activities that are either automatable today or will be within 12 months. That is a reallocation opportunity sitting unused.

Step 2: Separate task-based work from judgment-based work

Task-based work has a clear input, a defined process, and a predictable output: research the account, qualify the inbound lead, send the follow-up, update the CRM. Automating it frees your humans without diminishing what they do for the buyer.

Judgment-based work requires context, relationship history, and human decision-making: navigating a multi-stakeholder deal with competing priorities, reading the room when a negotiation shifts, knowing when to accelerate and when to slow down, deciding which accounts to prioritize when pipeline is underwater. The failure mode is treating both categories as equivalent, and either automating both or protecting both.

Step 3: Identify the growth that is currently out of reach

This is the steam engine question. Not what can AI do that we currently pay humans to do, but what growth exists that we cannot reach at our current human capacity and cost structure.

• SMB segments too small to justify a full AE
• Website visitors who arrive outside business hours and leave without engaging
• Trial users who drop off before reaching the feature that would convert them
• Accounts too small for dedicated CSM time but large enough to churn without attention

Those are the steam engine targets. New capacity, not efficiency gains on existing work.

Step 4: Build the reskilling bridge alongside the automation layer

Harvard Business School research on employee AI resistance traces the cause to framing. When AI is introduced as a threat to role identity, resistance follows predictably. The fix HBS prescribes is identity-compatible advantages: visible pathways showing employees how their expertise gains leverage from AI rather than getting replaced by it.

Kahlow's internal policy, where employees who successfully automate their own role get a higher-value position, is one concrete version of that. Made concretely and kept, that commitment is what separates an actual transition from a managed layoff.

Revenue leaders who navigate this well can state clearly, for every role on the team, which tasks AI will own and what that person will do with the time it frees up. Without both answers in hand, the automation becomes a quiet restructuring rather than a genuine capacity lift.

The close

The data is unambiguous: cutting headcount to fund AI investment does not generate returns. The organizations winning use AI to amplify human output, reach growth that was previously unreachable, and redeploy their best people to work that actually requires them.

The steam engine did not put the world's horses to sleep. It changed what the world needed from them, then opened a hundred industries horses alone could never have powered. Revenue leaders who move their best people toward the growth AI unlocks will pull ahead of the ones still defending a cost structure built for 2019.