Two Mandates, One Team | SBI Growth Advisory

Boards want AI results, and the quarter still runs on people

Every board now has an AI line on the agenda. A year ago, having a plan was enough; now it wants to see results. In our survey heading into 2026, 65% of CEOs said their customer-facing teams need AI-driven efficiency gains to hit their 2026 goals. The gains are already booked into the plan. The deployment is not. This year’s number depends not only on the sales team, but also on the operations team arming them with AI support.

Despite most teams expecting AI to help hit their goals, only one company in ten describes its AI in GTM operations as advanced or pervasive.¹ Fixing the distance between the plan and the reality lands on one team, RevOps.

65%

of CEOs say their customer-facing teams need AI-driven efficiency gains to hit their 2026 goals¹

1 in 10

companies describe their AI in GTM operations as advanced or pervasive¹

Revenue Operations is standing between two eras. Behind it sits the people era, when revenue scaled with headcount and the job was making sellers better through cleaner data, tighter territories, and sharper forecasts. Ahead sits the system era, when agents act on the data directly and a growing share of revenue comes from machinery the company runs. In between is the crossover, and that is where nearly every revenue team is standing right now.

The crossover period is hard because neither job lets go. Forecasts, territories, and pipeline still decide whether this quarter lands, and a team mid-transition needs that support more than ever. At the same time, the foundation the system era needs is being built now: the data the agents read, the workflows they run, the pricing inside them, and the governance that keeps them in the lines. RevOps is building the first new revenue source most companies have added in a generation, and the foundation decides who captures it. Both jobs are critical, but the stakes are not symmetrical. Miss a quarter and you lose a year. Miss the foundation and you spend the system era watching a competitor’s system compound on years of data you never captured, while your build waits for budget.

The outcome turns on three decisions a company controls: how it funds the two mandates, who it staffs them with, and the order it hands motions to the agents.

RevOps becomes a smaller, more technical team

When the crossover is over, RevOps will look different than before. In the system era, agents run many of the motions, and the consolidation work that fills RevOps calendars today, the reports, the dashboards, the quarterly readouts, is largely automated into the platform. What remains is the machinery and the people who run it. The RevOps function that emerges is smaller, more technical, agent-native, and continuously optimizing, closer to a product team than to the ops team most companies run now.

Side by side, the distance is plain.

RevOps moves from producing reports to producing revenue
	Today	Tomorrow’s system era
The work	Producing reports for human decisions	Operating the data, workflows, pricing, and governance that agents run on
The talent	Ops generalists, CRM admins, BI analysts	GTM engineers and agent operators
The scorecard	Revenue per rep, pipeline coverage	Revenue per rep for the sellers; AI run-cost share and revenue per GTM engineer for the system
The cadence	Quarterly planning, monthly reviews	Continuous optimization; capital allocation keeps the quarterly rhythm
The shape	A broad generalist team that scales with headcount	A small technical team that scales with what it builds

205%

growth in job postings for GTM engineers last year²

The talent row is the hardest move. A BI analyst rarely becomes an agent operator, and the market has noticed. Job postings for GTM engineers grew 205% last year.² With a new talent profile comes a new way to measure success. Revenue per rep keeps measuring the sellers who remain; revenue per GTM engineer joins it for the builders, now the scarce hire, and AI run cost as a share of revenue is new, the cost of running the motion on machinery rather than people.

That team takes years to assemble, and the system it runs takes years to mature, which makes stability its first requirement. Priorities that reset every few quarters kill the build. Who should hold the mandate is the subject of its own piece in this series; what matters here is that the build outlasts whoever holds the seat.

Agents take the simple motions first

share of each motion agents can run within three to five years; shares are illustrative

Velocity and PLG motions

Existing-customer support

Mid-market transactional

Complex enterprise

Velocity motions, where volume is high and cycles are short, are the first that agents run end to end, from outreach through close.

A product-led or velocity-heavy business can reach the system era inside three to five years.

agents run the motion end to end

Start where agents can run the whole motion; augment where people still build the deal.

Agent capability does not arrive everywhere at once. Agents take over simple, repeatable work first and complicated work later, so the order of adoption is predictable. Velocity motions, where volume is high and cycles are short, are the first that agents run end to end, from outreach through close. Inside the existing customer base, the support work around onboarding, adoption, and renewal prep follows close behind, because the tasks repeat and the data is already in-house. Complex enterprise selling sits at the far end. There, agents compress the work around a deal, the research, signal synthesis, follow-up, and forecasting, while people build the deal itself.

The same order tells you where to build. Start with a motion agents can run end to end. A whole motion trains the system faster than fragments of one, and it generates proprietary data a competitor cannot buy. From there, work down the list, piloting the deal-support work inside enterprise deals while a person still owns the outcome. Each step carries less risk than the last, because every motion the system runs teaches it something the next one needs.

How fast any of this moves depends on the motions a company sells through. A product-led or velocity-heavy business can reach the system era inside three to five years. A complex-enterprise business will spend longer in augmentation. The work resists handoff, and autonomy has to be earned, with internal controls and a thickening layer of AI rules deciding how much the agents are allowed to run. Neither path can be skipped. Both reward starting now, at the cheap end of the order.

The transition stalls unless the CEO drives it

RevOps’ mandate to change must come from the CEO and the board, because no team votes to automate its own budget. The standard prescription, fund the new mandate and starve the old one as it automates, lands on the team whose reporting work the agents automate away. That reporting work is where the team’s headcount, budget, and metrics all live, and every one of them rewards the old mandate. Asking a function to defund itself is how transitions stall in week one. So who drives the transition when the incumbent function has every structural reason to slow it?

Two design choices keep the mandate alive past the announcement:

The activation work gets its own budget line, separate from the line that runs the quarter, so the two mandates stop competing for the same dollars.
The crossover gets an end date, a stated point when carrying both stops, which turns an open-ended tax into a project with a finish.

The people part deserves more honesty than it usually gets. Some of today’s team converts. Analysts who know the data become its stewards, and the best process minds move to playbook upkeep and agent evaluation, checking what the machines did. Many do not convert, and pretending otherwise stalls the build as surely as the politics do. The defund itself runs as a sequence, one report at a time: automate the report, retire the report, move the budget. A managed-services partner can force the pace where internal conversion runs slow, since the partner has no attachment to the reports.

How long the migration takes depends on the starting point. Companies with mature RevOps run both mandates in parallel for 24 to 36 months and pay for the overlap. Companies with a thin function install the new one directly and skip the parallel cost. AI-native companies never built the old function and skip the migration entirely. Every established company makes the move eventually. The open question is whether they make it on purpose or get caught off guard.

System cost, not headcount, is what scales

PROOF CASE

Globant

System-era economics run on a different input. The clearest proof is in services, where revenue was always priced on people, so the moment AI breaks that link the margin jumps in plain view. Software margin was never labor-bound the same way, so its version of the break arrives later, but the mechanism is the same. Revenue grows with system cost rather than headcount, and the earliest hard numbers come from the companies furthest along. Globant, a digital engineering services firm, prices its AI Pods on token consumption rather than hours or seats, and reports gross margins of 45 to 60% on that work against a blended margin of approximately 38%.³ That pricing model is the tell: when the unit of cost is metered work instead of people, margin scales without hiring.

Managing the shift from headcount to system cost runs on the first of the new scorecard numbers: AI run cost as a share of revenue. Build-out spend sits apart in capital allocation, judged like any other investment. The math runs on a single motion: an outbound velocity motion carrying $40,000 a year in metered agent cost and producing $2 million in closed revenue runs at a 2% share, and the share falls as the motion scales because the cost meters on work rather than on heads. Treat the figures as illustrative. The pattern is what holds: a low AI run-cost share that falls as the motion scales. The exact dollars move with your motion, pricing, and vendor.

Run cost alone is not enough, because it only matters once the system carries real work. The crossover needs a second gauge: the share of revenue the system carries, the percentage of company revenue flowing through motions agents run end to end. It starts near zero and climbs one converted motion at a time. Read the two side by side each quarter. A rising system share at flat or falling run cost means the transition is working and the spend is sustainable. A rising share that holds only on ever-growing metered spend means it is not.

None of it is cheap. The crossover costs more than either era on its own, because both mandates draw money at once. The companies that handle it best treat the overlap as a priced, dated project rather than a permanent layer of spend, and they read the AI run-cost share as the gauge of whether the handoff is on schedule.

Run the crossover on purpose

Three moves put structure under the crossover:

Where to start

Set the end date and split the budget. Give activation its own line, hold decision support at what the quarter needs, and put a date on the overlap. The split stops the mandates from competing; the date stops the overlap from becoming permanent.
Sequence by motion. Start where agents can run end to end, velocity motions and existing-customer support, and let the adoption order set the rest. Each motion funds the next in proprietary data and proof.
Re-staff to the new scorecard. Hire the builder unit, GTM engineers and agent operators, name the transition paths for the analysts who convert, and measure the function on AI run cost as a share of revenue and revenue per GTM engineer.

Run the crossover as a managed project, and the operation gets leaner and cheaper to scale. The team that comes out is smaller and more technical, and its cost meters on the work the agents do rather than the heads you add, so output climbs without payroll climbing with it.

RevOps spent a decade consolidating data so people could act on it. The next one belongs to the team that activates it, with agents acting on the data directly and producing revenue the company can meter. The system era arrives either way. What you own when it gets here is the part you control.

Sources

SBI Q4 2025 CEO Survey, December 2025, n=118. 65% of CEOs say their customer-facing teams need AI-driven efficiency gains to achieve 2026 goals; 10.1% describe AI deployment in GTM operations as advanced (7.6%) or pervasive (2.5%).
Bloomberry analysis of 1,000+ GTM engineering job listings, 2024 to 2025 (205% year-over-year growth), as reported by eMarketer, 2026. ZoomInfo hiring data reported January 2026: postings doubled year over year for two consecutive years, with 3,000+ open LinkedIn postings in January 2026.
Martín Migoya (CEO, Globant), Q4 2025 earnings call, February 2026: AI Pods gross margins of 45% to 60% vs. a blended gross margin of approximately 38%; AI Pods are priced on token consumption rather than hours or seats.

Two Mandates, One Team: How AI Is Changing What RevOps Does