More revenue, fewer people: what AI means for MedTech companies

For a century, MedTech has grown the way most industries grow: revenue up, headcount up, roughly in lockstep. Gartner’s thesis, presented in a February 2026 webinar that became its highest-rated of the year, is that this correlation is weakening.¹ According to Deloitte’s survey of 100 MedTech executives (May 2026), 57 per cent still use AI mainly for individual productivity, and 90 per cent admit that fewer than 15 per cent of their processes have been materially transformed.² The advantage will go to organisations that stop treating headcount as the underlying driver of results and start deliberately setting the human-to-AI mix inside each process.

Researched and drafted with AI assistance; reviewed, fact-checked and edited by Primož Verbič.

In brief

• Gartner’s 2026 model says the ratio of people to AI inside each process is now a decision, not a consequence of growth; the correlation between revenue and headcount is weakening.
• MedTech is early: 57% of companies still use AI for individual productivity, 90% have materially changed fewer than 15% of processes, and only 6% have redesigned processes so people and agents share the work (Deloitte, May 2026).
• Hierarchies flatten, but the number of direct reports per manager falls before it rises, and the damage lands on beginners: 22–25-year-olds in AI-exposed jobs are 19% below trend (Stanford, August 2026).
• Producers reorganise by therapeutic area with regional field presence; RA/QA becomes the AI governance hub. Distributors automate the back office and survive only by owning a knowledge base on product use.
• Eight measures make the change sustainable, from redesigning processes rather than jobs to ISO 13485 validation of internal agents and commissions that reward the new logic.

The model behind that thesis is simple, and it is laid out below. Five stages of enterprise AI investment: Everyday AI, Augment, Semiautonomous, Autonomous and Agency. From the first stage to the last, the share of work done by humans falls from 100 per cent to 20 per cent, while the size of the process change rises from small and incremental (Amplify) through moderate (Recalibrate) to significant (Reset). Gartner’s point is not that everyone should sprint to the highest stage. It is that the position on this spectrum should be chosen per process, with eyes open.

This article offers MedTech leaders three things. It describes how producers and distributors are actually organised today. It maps Gartner’s five stages onto the work we do and, drawing on the latest organisational research, predicts how structures will change. And it sets out the measures that separate a sustainable transition from an expensive one.

Where MedTech stands today

The honest answer is: mostly at the early stages of the model, with ambitions well above that. Deloitte’s survey shows that only 21 per cent of executives believe today’s productivity-focused use of AI reflects their ambition, and that 45 per cent already have agentic AI in production. But only 36 per cent think they have optimised processes around AI, only 7 per cent of those already using agents describe them as fully embedded and audit-ready, and only 6 per cent have redesigned processes so that people and AI agents share the work.² To see why the gap is so wide, look at how the two halves of the industry are organised.

Producers

A mid-sized European or US manufacturer is a functional pyramid: R&D and clinical research, regulatory affairs and quality, operations and supply chain, and a commercial organisation that is almost always the largest block of people. Commercial itself splits into direct territory reps (often with OR or lab presence), clinical or application specialists, key account managers (KAMs) for hospital groups, tender and public-procurement teams, marketing and customer care. Country subsidiaries replicate the pattern with two or three management layers between the field and the regional director.

Coordination is the glue of this structure. Product managers relay clinical evidence to reps; area managers relay forecasts upward; RA/QA staff relay standards downward. The 2026 trends review by consultancy ZS Associates notes that many large MedTech organisations are decentralising into more autonomous business units while their customers, ambulatory surgery centres and hospital networks, buy across categories, which forces investment in unified commercial models, shared customer data and coordinated contracting.³ In other words, the structure is already bursting at the seams before AI enters the picture.

Distributors

A specialist distributor (the model that dominates continental Europe, Latin America and much of Asia) is leaner and flatter, but no less labour-coupled. Its value sits in four places: inventory and consignment management, including sterile trays and loaner instruments; a sales force with close relationships with doctors and nurses, covering the portfolios of several represented manufacturers (principals); a tender and registration desk that knows local procurement law; and an order-to-cash back office that manually reconciles hospital purchase orders, delivery notes and rebate claims.

Headcount follows lines and accounts almost mechanically. Add a manufacturer, add a product specialist; add a region, add a rep and a driver. The margin that pays for all of this is under pressure from both sides: hospital supply chain leaders now buy on total cost of ownership and effectiveness data rather than the preferences of clinical staff alone,⁴ and joint public procurement and purchasing consortia increasingly require structured data and electronic document interchange (EDI) rather than free-text PDFs.⁵ A distributor that cannot meet those data requirements is not slow; it is excluded.

Mapping Gartner’s spectrum onto MedTech work

Gartner’s five stages are generic. Below is what each looks like when the process is a MedTech process. For each stage I describe what it means for a producer, what it means for a distributor, and where the industry is today.

Two things stand out from the mapping. First, the stages are not evenly hard. Moving from Everyday AI to Augment is a tooling decision; moving from Augment to the Semiautonomous stage is an organisational one, because it is the first point at which an AI output can reach a customer, a regulator or a patient record without a human touching it. That is where the size of the process change moves from Amplify to Recalibrate, and exactly where most MedTech companies get stuck. Second, the Agency stage, where AI acts with delegated authority, is not a distant abstraction for this industry. A replenishment agent that places consignment orders, or a tender agent that submits a compliant bid, is already technically feasible. What is missing is clearly assigned responsibility for oversight.

What the research says about AI and organisational structure

We often hear that AI will eliminate middle management. The research of the past eighteen months says: yes, partly, but not immediately, not everywhere and not for free. Those details are exactly what leadership teams need.

Hierarchies do flatten, and managers lead more people

Shan and Zhu, in a March 2026 paper using 7,681 US firms across three decades, find that AI exposure is associated with systematic restructuring: fewer hierarchical ranks and more direct reports per manager.⁶ Korn Ferry’s survey of 15,000 professionals found 41 per cent saying their organisation had cut management layers in the past year, and practitioners report the average number of direct reports per manager rising from around eight in 2013 to twelve in 2025.⁷ In MedTech terms: the area manager whose main job is aggregating forecasts and relaying pricing exceptions is the most exposed job in the company.

“a lot of management became workflow babysitting, and AI agents are very good babysitters”

— Mark Vena, CEO and principal analyst, SmartTech Research, in Fortune, June 2026⁸

But not evenly, and not from the top down

A 2026 model of generative AI in knowledge hierarchies by Xu and colleagues shows that the number of direct reports per manager falls before it rises: early automation reduces the need for first-level staff while demand for senior expertise stays flat or rises, because someone has to validate fallible AI output. Only as the technology matures and juniors upskill does the structure genuinely flatten.⁹ This matches what MedTech is living through. The first thing agentic AI does in a quality function is generate more work for the senior reviewer, not less.

The damage is showing among beginners

Stanford’s Digital Economy Lab, using payroll data from ADP, a provider that processes pay for more than 25 million US employees, through June 2026, finds no economy-wide displacement, but employment of 22-to-25-year-olds in AI-exposed occupations now sits 19 per cent below where it would be had it tracked their less-exposed peers. The gap has widened every month since it was first documented, and it operates through reduced hiring rather than layoffs.¹⁰ Economists call this technological change in favour of the experienced: AI substitutes for entry-level execution while complementing expert judgment. The trouble, as a 2026 paper on knowledge transmission points out, is that expertise is not a given. If juniors never do the routine work, the future supply of the senior judgment AI depends on withers with it.¹¹ Deloitte’s MedTech survey names the same risk directly, a “broken career ladder” in quality, regulatory and clinical research.²

Layoffs without process redesign do not pay

Gartner’s own data is the most sobering. In a survey of 350 executives at companies with over a billion dollars in revenue, 80 per cent of those piloting AI or autonomous technology reported workforce reductions, but the cuts happened regardless of whether the technology was generating returns.¹² Separately, Gartner projects that through 2028 AI investment can produce a net headcount increase inside an enterprise, as high as 30 per cent in some units, typically under 10 per cent for knowledge workers overall, because roles are created as well as removed.¹³

“Looking only at layoffs is shortsighted in terms of getting value from AI”

— Helen Poitevin, VP Analyst, Gartner, in Fortune, May 2026¹²

CFOs, meanwhile, have already decided. Gartner’s 2026 budget benchmarks show expected headcount growth collapsing from 6 per cent to 2 per cent while three-quarters of finance leaders raise technology budgets, which one Gartner finance analyst described as “a structural pivot from labor expansion to optimization”.¹⁴ The money is moving whether or not the operating model is ready.

The shape of a MedTech organisation in 2030

Put the research together with the spectrum and a picture emerges. It is not the same picture for producers and distributors.

Producers

Three shifts define the producer of 2030. The commercial pyramid shrinks into teams organised by therapeutic area or customer segment; field presence stays regional, and agents do the retrieval, drafting and coordination that area managers and marketing associates do now. The rep does not disappear; the rep becomes a clinical-economic consultant whose back office is a set of agents. Alex Wakefield, chief revenue officer at AcuityMD, put the direction bluntly in Med-Tech Insights’ 2026 predictions: “Commercial teams will operate like technology companies.”¹⁵

Regulatory and quality change in substance, not in size. In a world where an agent influences a complaint investigation, a CAPA process (corrective and preventive action) or a labelling decision, RA/QA becomes the function that validates, logs and controls agents, which is why Deloitte finds that orderly data governance and unified, standardised core processes are prerequisites for scaling AI in a sector that has historically been highly decentralised: autonomous business units, country subsidiaries, each with its own tools.¹⁶ Finally, the team becomes more experienced and, at the same time, deliberately rejuvenated: fewer people in the middle, more expertise at the top, and an explicit pipeline to grow the next cohort of experts because the market will no longer grow them on its own.

Distributors

For distributors the spectrum splits the business in two. The back office, order desk, tender monitoring, consignment replenishment and claims reconciliation, is a near-perfect candidate for the Autonomous and Agency stages; it is rules-based, data-rich and invisible to the customer. Becker’s reporting from hospital supply chain leaders in 2026 is instructive: the solutions that are actually working are the unglamorous ones, removing repetitive work and improving visibility, not autonomous purchasing decisions.¹⁷ The field changes far less. In implantable and procedural categories the rep is part of the product, and delivery reliability plus rep quality remain decisive in hospital selection.⁴

For the distributor the danger is different. If it uses AI only to cut its own costs, it becomes cheaper but no more necessary; a manufacturer with its own agents can supply hospitals without it. The distributor stays indispensable only if it invests the time and money it saves in what neither side has. To the hospital it offers product onboarding with professional staff training, monitoring of use, and data on improved effectiveness for the patient, the staff and the organisation, and ultimately for the whole hospital network and the health budget. To the manufacturer it offers continuous feedback on needed product improvements and stable sales growth. Whoever builds that stays in the chain; whoever only cuts costs drops out of it.

What makes the AI transition sustainable in MedTech

A sustainable change here means three things at once: that the returns are real, that the organisation still works in 2032, and that the regulator does not stop you. The evidence points to eight measures. Tick the ones you already have in place.

• Redesign the process, not the job.Only 11 per cent of MedTech executives say AI enablement is integrated into onboarding and upskilling with clear accountability. Adding an agent to a role built for a pre-agent world produces the Deloitte pattern of AI outputs used informally with no updated SOPs.² Start from the process map, decide the human-AI split per step, then rewrite the job descriptions.
• Assign responsibility for every process.Before an agent goes live, name who owns the result if the agent fails, who approves changes and who decides when a case falls outside its design. A centre of excellence several layers up is documentation, not accountability.²
• Measure results, not headcount.Gartner's guidance to sales leaders is to decouple metrics from headcount assumptions, identify which activities drive results and set productivity baselines independent of FTEs.¹⁸ In quality, that means measuring cost per resolved complaint and days to close, not how many AI pilots you have launched.
• Deliberately preserve the career ladder and the skills that come with it.Stanford's data and the knowledge-transmission literature both say the market will stop growing your future experts.^10,11 Reserve a share of routine RA, QA and field work for juniors under supervision, involve them in agent oversight, and treat the onboarding of beginners as an investment rather than a cost to be optimised away.
• Set up oversight the regulator will recognise.The Digital Omnibus, in force since 27 July 2026, moves high-risk obligations for AI embedded in MDR/IVDR devices to 2 August 2028; medical devices remain in Annex I with no carve-out, and the Article 50 transparency duties applied from August 2026.^19,20 Internal agents that touch regulated records follow a different route: ISO 13485 (clause 4.1.6) requires validation of any software used in the quality system before use and after changes, which notified bodies check at every audit. So aim now for validation, audit trails, change control and post-deployment monitoring; ISO 42001 is a usable framework for AI governance.
• Fix the data first, then scale agents.Only 22 per cent of MedTech executives describe their data as broadly searchable, reused and documented; 61 per cent say shared tools exist but are used differently across functions.² Know where the data is sound enough to run, where to build first and where to wait.
• Choose the stage per process, and write it down.The size of the process change, Amplify, Recalibrate, Reset, is a decision, not a consequence. A complaint-intake process may belong at Semiautonomous; surgeon education may belong at Augment for a decade. An internally agreed rule ensures that your strategy sets the direction, not the latest demo from an AI-tool vendor.
• Commissions and bonuses must reward the new logic, not the old.At Davos in January 2026, The Permanente Medical Group's CEO described AI deployed where the strain was measurable, physician retention and burnout, rather than as a showcase, and noted that in fee-for-service settings efficiency can conflict with revenue while in value-based settings it aligns.²¹ The same logic applies to a distributor rep paid per case or a producer paid per unit shipped.

0 of 8 measures in place.

“We don’t make profit by doing more. We make profit by doing better.”

— Dr Maria Ansari, CEO, The Permanente Medical Group, at the World Economic Forum, January 2026²¹

Siemens Healthineers’ Amira Romani added a related thought on the same panel: an AI-enabled imaging protocol that cut MRI scan times from around ten minutes to two cleared three-month diagnostic backlogs in Nordic customers, and remote reading lets radiologists cover several departments at once. Philips’ Marnix van Ginneken described AI-guided handheld ultrasound that cuts the training time for field screening staff from months to weeks, while acknowledging that regulatory systems were not designed for adaptive AI.²¹ The pattern in both is the one Gartner is describing: scarce expertise redistributed rather than replaced, and the organisation redesigned around that redistribution.

The choice in front of us

Revenue growth and headcount growth in MedTech will no longer go hand in hand, whether we like it or not. CFOs already assume it in their budgets. The only question is how the change arrives. The first path is deliberate: for each process we decide what people do and what AI does, and back it with oversight, data and a preserved career ladder and the skills that come with it. The second path is the one Gartner is already observing: layoffs on the promise of AI before AI has delivered anything, leaving a hollow middle of the organisation and a sharply reduced lower level, which in turn means today’s experts have no successors.

A mid-sized producer should immediately pick two processes, one commercial and one in quality, map them to the spectrum, and rebuild them end to end with accountability and metrics set before deployment. A distributor should decisively automate the back office and reinvest every freed hour into the knowledge base on product use that makes it indispensable to both hospital and manufacturer. Neither requires a Reset across the whole company. Both require deciding, in writing, which stage each part of the business is meant to be at.

A note on scope. This article addresses organisational AI: the agents and tools that run MedTech companies. AI inside the medical device itself (software as a medical device, SaMD; AI-enabled diagnostics) follows the MDR/IVDR conformity route and the 2028 AI Act timeline referenced above, and deserves its own treatment.

Common questions

Is an internal AI agent a medical device under the EU AI Act? No. The AI Act’s 2 August 2028 deadline applies to AI that is part of the device itself. An agent that helps triage complaints or maintain documentation is quality-system software under MDR and ISO 13485 clause 4.1.6, and must be validated before use and after changes.

Which of Gartner’s five stages should a MedTech company aim for? There is no single right stage. The point is to choose per process: complaint intake may belong at Semiautonomous, surgeon education at Augment for years. The transition that matters most is from Augment to Semiautonomous, because that is the first stage where AI output reaches a customer or a regulated record without a human touching it.

Will AI reduce the field sales force in MedTech? Far less than the back office. In implantable and procedural categories the rep is part of the product; delivery reliability and rep quality remain decisive in hospital selection. The most exposed commercial role is the area manager whose work is aggregating forecasts and relaying pricing exceptions.

How can a distributor avoid being bypassed by manufacturers’ own AI? By building what neither hospital nor manufacturer has: a knowledge base on product use, meaning onboarding and staff training, monitoring of use and effectiveness data for the hospital, and continuous product feedback and stable growth for the manufacturer. A distributor that only automates its own costs becomes cheaper, not more necessary.

Sources

1. Gartner, Beyond the Hype: AI’s Impact on Headcount and Business Value, webinar, 25 February 2026; and AI’s Impact on Jobs: Decoupling Headcount From Revenue, 2 January 2026.

2. Deloitte Center for Health Solutions, Medtech’s agentic AI moment: Turning ambition into enterprise advantage, 30 July 2026 (survey of 100 MedTech executives, May 2026).

3. ZS, Medtech trends 2026: From AI strategy to execution, February 2026.

4. SLR Medical Consulting, Medical Device Distribution: How the Supply Chain Works, July 2026; and Medical Device Industry Trends Reshaping 2026.

5. MedStrato, GPO AI Procurement Automation 2026, May 2026.

6. Shan, G. and Zhu, F., AI Exposure and Organizational Structure, SSRN, 22 March 2026.

7. Forbes, AI Flattening Organizations Is The Latest Chapter In A Continuing Story, 21 May 2026, citing Korn Ferry’s 2025 Workforce survey.

8. Fortune, AI agents are flattening corporate hierarchies, 9 June 2026.

9. Xu, F. et al., Generative AI and Organizational Structure in the Knowledge Economy, arXiv, 2026 revision.

10. Brynjolfsson, E., Chandar, B. and Chen, R., Canaries in the Coal Mine, August 2026 update, Stanford Digital Economy Lab.

11. Anonymous working paper, Automation, AI, and the Intergenerational Transmission of Knowledge, arXiv, 2026.

12. Fortune, AI isn’t paying off in the way companies think, May 2026, reporting Gartner’s survey of 350 executives.

13. Poitevin, H. and Suda, N., Gartner, Why Your Headcount Strategy Matters More Than AI Downsizing, 23 January 2026.

14. The CFO, 2026 Budget Analysis: CFOs Pivot to AI and Tech Over Headcount, 17 February 2026, reporting Gartner benchmarks.

15. Med-Tech Insights, MedTech industry predictions for 2026, December 2025.

16. Deloitte, Three Key Trends Likely to Shape Medtech in 2026, January 2026.

17. Becker’s ASC Review, Supply chain updates 2026, August 2026.

18. Gartner, Sales Operations for the AI Era, July 2026; and AI Reveals Why Sales Productivity Metrics Are Broken, May 2026.

19. Legalithm, AI Act and medical devices: deadlines and Omnibus tracker, 30 July 2026 (Regulation (EU) 2026/1744).

20. RAPS, EU Commission drafts guidelines on classifying high-risk systems under the AI Act, June 2026.

21. Forbes, AI And The Workforce Reset In Healthcare, 3 March 2026, reporting the Marsh panel at the World Economic Forum Annual Meeting, 21 January 2026.

Model concept and human/AI percentages: Gartner, “Decoupling of headcount to revenue”, 2026. All quotations are attributed to the named speaker as reported by the linked publication; all other source material is paraphrased.