Gartner expects 60% of organizations to run smaller software engineering teams at scale by 2029, up from 15% in 2026. US tech employers surveyed by Experis rank professionalism and critical thinking above AI literacy. If your 2027 headcount plan still counts engineers the way it did in 2024, it is planning for a team shape that is disappearing.
On September 22, 2026, Experis, part of ManpowerGroup, published its fourth quarter Tech Talent Outlook. The US Net Employment Outlook for tech employers came in at 37%, down 10 points from both the previous quarter and the same period last year. Still, 53% of US tech employers plan to increase staffing in the fourth quarter, 30% expect to hold steady, and 16% expect reductions.
Kye Mitchell, President of Experis US, summarized it this way: "U.S. tech hiring is moderating, but this is a market getting more deliberate, not pulling back."
Deliberate is the right word, and it describes more than the pace of hiring. It describes the shape of the teams companies are hiring into. They are getting smaller, and the people on them are expected to bring more judgment than before.
A disclosure before we go further. ManpowerGroup is a staffing company, and so are we. Both of us benefit when companies rethink how they staff. We have tried to lean on the analyst and academic evidence in this piece and to be clear about where each number comes from.
What Is Actually Changing About AI Team Size?
More software engineering work is moving to tiny teams of four or five people built around AI tooling. Gartner predicts that 60% of organizations will adopt smaller software engineering teams at scale by 2029, up from 15% in 2026, and calls the change a restructuring rather than a cost-cutting measure.
Gartner's July 2026 release describes today's tiny teams as typically four to five members, with some as small as two or three. Its recommended composition is lean: a product manager, a user experience or agent experience designer, and at least one AI-native software engineer.
Aliyah Camacho, a Principal Analyst at Gartner, was explicit about the motive: "Tiny teams are not a cost optimization tactic. This is a restructuring of teams to best take advantage of both human and AI capabilities and strengths." She also argued that AI is "fueling the demand for more software engineers, not fewer." Those two statements fit together. Organizations may run more teams, each smaller, rather than fewer engineers overall.
McKinsey's State of AI survey, published August 25, 2026 and based on 1,719 respondents in 97 countries, shows how far expectations are running ahead of reality. Among respondents at organizations using AI, only 14% reported that AI contributed to a decline in overall workforce size over the past year. Across the survey, two-thirds reported little or no AI-related change in total employment. Yet 39% expect AI to decrease their organization's headcount over the next year. The change is coming, but it hasn't arrived yet, which means 2027 is when many companies will make these decisions.
Why Are the Teams Getting More Senior?
Because when a team of four is doing what a larger team used to do, every seat has to exercise judgment, and judgment is what experience buys.
The Stanford Digital Economy Lab's August 2026 update, by Erik Brynjolfsson, Bharat Chandar and Ruyu Chen, using ADP payroll data through June 2026, found that employment among workers aged 22 to 25 in highly AI-exposed occupations now stands about 19% below where it would be had it kept pace with similarly aged workers in less-exposed jobs. That gap was 15% a year earlier. Experienced workers show no comparable gap. The authors found that employment was increasing among experienced workers in occupations that rely on tacit knowledge acquired through practice, mentorship, and repeated exposure to real-world situations. They describe these as descriptive patterns, not causal estimates of the effect of AI, and caution that caution is worth keeping.
The skills employers say they want point the same way. Among US tech employers in the Experis survey, the most sought-after skills were Professionalism and Work Ethic at 44%, Critical Thinking and Problem Solving at 39%, Adaptability and Willingness to Learn at 37%, and Communication, Collaboration and Teamwork at 35%. AI Literacy was the most sought-after technical skill at 34%, with AI Modeling and App Development close behind at 33%. The wider survey covered 4,258 tech and IT services employers across 42 countries and was collected from July 1 to 31, 2026, but the release does not disclose the US sample size.
KPMG's second-quarter 2026 AI Pulse, a survey of 204 US leaders at companies with $1 billion or more in revenue, found a similar trend at the executive level: 54% agree that strong social and interpersonal skills are more important for career success than strong mathematical or technical skills.
Put plainly, the AI handles more of the typing. What remains is deciding what to build, spotting when the output is wrong, and explaining trade-offs to the people who own the business outcome. We covered where these engineers come from, and who else is bidding for them, in who you are actually competing with for AI engineers.
Is a Smaller Team Actually Cheaper?
Not automatically, and planning as if it were is a costly mistake.
Gartner predicted in June 2026 that by 2028, AI coding costs would overtake the average developer's salary, driven by rising token consumption and a shift to consumption-based licensing. Nitish Tyagi, a Senior Principal Analyst at Gartner, noted that "most organizations still lack the maturity and frameworks to effectively measure cost versus business impact."
That changes the arithmetic of a small team. If Gartner's forecast holds, a four-person team with heavy agent usage could cost as much as a larger team, with the difference shifted from payroll to a tooling line that finance may not be tracking. We wrote about that line item in why your AI bill is exploding. For staffing purposes, the point is simpler: budget people and tokens together as a single team cost, or you will make headcount decisions based on half the data.
There is also a pipeline cost that does not show up right away. Gartner warns that by 2028, organizations that rely on AI to cut junior roles will hollow out their own software engineering talent pipeline. Camacho named the consequences: inhibited knowledge transfer, a restricted internal pipeline, and recruitment limited to more expensive and competitive senior roles. A company that goes all senior in 2027 is choosing to buy every future senior engineer on the open market. We made the full case in the hidden cost of cutting junior developers.
What Does a Small AI Team Actually Need?
Gartner's composition is a good starting frame. In our view, small AI teams need to cover four responsibilities, regardless of their job titles.
Someone who owns the outcome. A product owner who can say what success looks like in business terms and kill work that does not move it. On a four-person team, this cannot be a part-time role borrowed from another group.
Someone who owns the architecture. A senior engineer accountable for how the system fits together, how data flows, and where the model is and is not trusted. This is the seat most worth hiring into your own organization, because the context it accumulates is the context you least want to lose. Our forward deployed engineer post covers a close cousin of this role.
Someone who owns quality and evaluation. When agents write a growing share of the code, reviewing it, testing it, and measuring whether the model is behaving is a full job, not a side task.
AI-native builders. Engineers fluent enough with agents and models to multiply the team's output, and senior enough to know when the output is wrong.
A team of four covering those responsibilities is not a team of four generalists. It is a team of four specialists, each of whom covers a wide range of areas. That is why the hiring bar goes up as the headcount comes down.
How Should You Build Your 2027 Staffing Plan?
Plan by team, not by headcount. Start from the number of product areas that need a dedicated small team, then staff each one. A plan that starts from "we need 40 engineers" will reproduce the old shape.
Budget people and AI tooling as one line. Give each team a combined budget covering salaries and token consumption, and review it quarterly. Gartner's salary crossover prediction is a forecast, not a certainty, but the direction is clear enough to plan for.
Hire the architecture seat in-house. The senior person who owns the system and its context should be yours. That is where turnover hurts most.
Keep a junior lane open on purpose. One apprentice seat per two or three teams, paired with a senior engineer, costs little and protects the pipeline Gartner warns you are about to hollow out.
Use flexible capacity for the builder seats. The AI-native builder and quality seats are areas where a dedicated external team can quickly add senior-level capacity without having to bid in a heavily contested part of the US market. A dedicated AI team in an overlapping time zone works here because small teams depend on fast, frequent conversation. Our AI staffing models guide compares the engagement structures in more detail.
Fund the upskilling, do not assume it. KPMG found upskilling and reskilling remain the top workforce strategy for 65% of the large-company leaders it surveyed. That is a statement of intent. Turning it into capability takes time, budget, and structure, which is what our AI fluency program is built to provide.
Common Questions About Staffing Smaller AI Teams
Are AI engineering teams really getting smaller?
The direction is clear, though the change is early. Gartner predicts 60% of organizations will adopt smaller software engineering teams at scale by 2029, up from 15% in 2026, with typical teams of four to five people. McKinsey found only 14% of respondents at AI-using organizations saw an AI-driven workforce decline last year, while 39% expect one next year.
Does a smaller AI team cost less?
Not necessarily. Gartner predicts that by 2028, AI coding costs will overtake the average developer's salary as token consumption rises and licensing shifts to consumption-based pricing. A smaller team can shift costs from payroll to tooling rather than removing them. Budget people and AI tooling together as a single team cost to see the real number.
What skills matter most for AI engineering teams in 2027?
Judgment skills lead the list. Among US tech employers in Experis's fourth quarter 2026 outlook, professionalism ranked highest at 44%, followed by critical thinking at 39% and adaptability at 37%. AI literacy was the top technical skill at 34%, with AI modeling and app development at 33%.
Should we stop hiring junior engineers if AI teams are getting more senior?
No. Gartner warns that by 2028, organizations relying on AI to cut junior roles will hollow out their own talent pipeline, leaving them to recruit only expensive, contested senior engineers. A small, deliberate junior lane, with each apprentice paired with a senior engineer, is cheaper than buying every future senior on the open market.
If you are building a 2027 plan around smaller, more senior teams and want to decide which seats belong in-house and which can flex, talk to the Golabs team. We will start with your team structure, not a headcount number.