Only 15% of Life Sciences and Health Care Firms Rate Their Talent Ready for AI. Train the Managers First.

TLDR: Life sciences should spend its first enablement money on the managers who redesign the work, because the sector’s own leaders rate talent as their least prepared area for adopting new technology.

Life sciences nearly doubled access to artificial intelligence (AI) tools, and only a quarter of users rely on them daily

In its State of AI in the Life Sciences & Health Care Industry, published in March 2026, the Deloitte AI Institute reports that workforce access to AI in the sector grew from under 35% to 60% of workers in one year¹. The same survey finds that among the workers who have access, only 24% use AI in their daily workflow more than 60% of the time¹. Access has spread across most of the workforce, while daily habit has formed in about a quarter of it.

Deloitte surveyed 3,235 leaders, from director level to the C-suite (chief-level executives), in 24 countries between August and September 2025, split equally between information technology (IT) and line-of-business roles, all from organisations with at least one AI implementation in daily use¹. The life sciences and health care cut, which this article relies on, covers 378 of them. These are leaders assessing their own organisations, at companies Deloitte describes as being on the leading edge of AI.

Production tells the same story. Deloitte finds that 28% of these life sciences and health care organisations have moved 40% or more of their AI experiments into production, and that 57% expect to reach that level within three to six months¹. Deloitte’s pure life sciences survey, the 2026 Life Sciences Outlook², published on 9 December 2025, puts it more bluntly. Of 280 biopharma and medtech C-suite executives in the United States, France, Germany, Switzerland, the United Kingdom, China and Japan, only 22% said they have successfully scaled AI, and just 9% reported significant returns². On integrating AI tools into daily workflows, 14% report full implementation and another 40% are working towards it². Of the executives transforming their organisations for AI-driven ways of working, the Outlook notes that none report having fully completed the journey².

Put together, these figures move the bottleneck. A firm that has given six in ten staff a sanctioned AI tool has solved procurement. Still open is the question of which tasks those staff now do differently, who decides that, and who answers for the result when an output is wrong. Those are questions about how work is organised, and in a regulated company the people who organise work are its managers.

Talent is the least prepared of five AI-readiness areas

Deloitte asks leaders to rate their preparedness to adopt AI broadly across five areas. In life sciences and health care, the share rating themselves highly or very highly prepared is 34% for strategy, 34% for data management, 32% for technology infrastructure, 28% for risk and governance, and 15% for talent¹. Talent sits at the bottom by a wide margin, and it slipped by one percentage point on the previous year, as did data management. Strategy rose four points, although it remains well below the cross-industry average of 42%. Risk and governance rose eleven points, the largest gain of the five¹.

Strategy and governance, Deloitte writes, can advance more quickly than infrastructure and talent “because they depend primarily on executive decision-making and policy development”¹. Strategy and policy are decided at the top and take effect once approved. Talent readiness moves only when thousands of people change how they spend their working day, and that change happens team by team.

The leaders say as much themselves. According to the report, insufficient worker skills are the biggest barrier¹ to integrating AI into existing workflows. That sentence is easy to read as a call for more training of the same kind, yet Deloitte’s own talent data, set out below, point to the roles that redesign work.

Two caveats apply to the 15%. Deloitte’s industry group is life sciences and health care together, so providers sit alongside pharma, biotech and medtech in the 378. And the figure is a self-assessment by senior leaders, which makes it a measure of their confidence in their own workforce. On that reading, the people who sign off the AI budget rate the human side of adoption as their weakest asset.

The talent budget goes to fluency for everyone and to specialist hires

Asked how they are adjusting talent strategy because of AI, the largest group of life sciences and health care organisations, 53%, is educating the broader workforce to raise AI fluency¹. Next come designing upskilling and reskilling strategies at 48% and hiring specialised talent to drive AI initiatives at 39%. Further down, 29% are combining or reimagining organisations around new work patterns and 28% are redesigning career paths. Deloitte sums it up: fewer than half are making significant adjustments to talent strategy, and “far fewer are re-architecting roles, workflows, and career paths”¹.

The job data confirm the pattern. In the same survey, 89% of life sciences and health care companies have not redesigned jobs around AI capabilities¹. The Outlook adds that 29% of biopharma leaders and 31% of medtech leaders plan to use AI tools or training to improve workforce productivity in 2026², which again treats training as a lever applied to everyone at once.

The two most popular responses, a fluency course for the whole workforce and a hiring push for specialists, both pass over the layer that decides how the work actually changes. A fluency module teaches a medical information associate how to write a better prompt. A data scientist hired into a central AI team builds a better model. The decision that the first draft of a standard response letter now comes from the tool, that a named reviewer checks it against the approved label, and that the team’s weekly target moves accordingly, sits with the medical information manager. With 89% of firms yet to redesign a single job around AI, few of those managers have been asked, or trained, to make it.

The result is the profile in the access figures: 60% of workers hold a sanctioned tool and 24% of them rely on it daily. The 24% daily-use figure fits a pattern in which staff given a tool while their tasks stay the same keep the old process. If the gap sits in the process, more of the same fluency course leaves it open.

Redesigning the work is a manager’s job

Deloitte names the shift directly. As front-line jobs become more automated, it writes, supervisor and managerial roles will likely shift toward orchestration of human-AI teams¹. Many firms are already thinking about what that means for structure: 47% of life sciences and health care companies have considered pod-based or non-hierarchical models, yet only 14% have moved to such models to a great or maximum extent¹.

The report’s own example is a clinical operations manager or care team lead who has always managed patient flow with clinical judgement, standard protocols and retrospective reports, and who now works alongside systems that generate real-time risk scores, care pathway recommendations and capacity forecasts. Deloitte poses the questions that manager now faces: “When should they override the AI? How do they explain their decisions?”¹

The same questions reach regulated functions in pharma and medtech. The three cases below are illustrations drawn from the Kainjoo Institute’s reading of these functions, and each describes a decision a manager would own. A pharmacovigilance (drug-safety monitoring) case-processing lead has to decide which steps of case intake an AI tool may pre-populate and which a qualified person must still assess. A medical-legal-regulatory (MLR) review owner has to decide whether an AI-drafted claim summary can enter the review cycle and who attests that it matches the reference. A quality assurance manager working under Good Practice (GxP) quality rules has to explain to an auditor why an AI-assisted deviation assessment was accepted. In each case the answer typically becomes a written procedure with a named signatory, drafted by the manager.

A kainjoo.life analysis of case managers and discharge prediction AI³ shows why that layer matters: in the final 24 hours before discharge, case managers’ estimates carried half the error of the tool’s, and the analysis concludes that such a tool earns trust when it informs the case manager’s plan at the moment of decision. It echoes the Kainjoo Institute’s earlier piece on AI adoption⁴, which concluded that technology was never the bottleneck, and that people and culture were. The 2026 data now show where in the organisation those people sit.

Managers split almost evenly on readiness, and confidence follows training

The closest available evidence on manager readiness is cross-industry and comes from the United States. On 28 July 2026, Indeed published a survey run with YouGov, fielded between 22 May and 2 June 2026, covering 1,301 respondents: 1,001 job seekers and 300 hiring decision-makers at organisations with more than 500 employees. Among those 300 decision-makers, 45% of managers feel equipped, while 43% do not⁵, to lead what Indeed calls AI-native talent.

Indeed defines an AI native as someone “highly skilled and comfortable with AI technologies” who will “fundamentally default to AI to design, execute and scale their workflows end to end”⁵. Fewer than one in six workers describe themselves that way, and 52% of workers say they are not getting the AI training they need⁵. The question Indeed asked managers is therefore narrower than general AI readiness: it is whether they can lead people whose working method is built around AI.

Two further findings matter for budget holders. Among managers who already lead AI-native staff, nearly nine in ten say they feel equipped⁵. And 88% of employers that already recruit AI natives provide manager training, compared with only 8% of employers that do not⁵. That is an association, and the two groups differ in more than training, since one of them is already hiring for AI skills. It does show that the firms furthest along have chosen to train their managers, and that exposure and preparation go together.

The finding transfers to life sciences for a simple reason. By giving 60% of workers sanctioned AI access, the sector is creating AI-reliant staff inside its own teams, even as it recruits few AI natives. Each of those teams has a manager whose decisions determine whether the work is redesigned around the new habit. A manager layer that splits almost evenly on readiness, in a sector that rates its talent the weakest of five readiness areas, is where the next enablement euro or franc earns the most.

A five-question check shows which managers can already lead AI-changed work

The following check is a working tool developed by the Kainjoo Institute for one function at a time. It is derived from the role changes described above and is a practical diagnostic, unvalidated as a survey instrument. Score each question 0 (no), 1 (partly) or 2 (yes) for one manager and their team.

1. Task map. Can the manager name which of the team’s recurring tasks AI now performs, which it assists, and which stay entirely with qualified people, such as GxP-relevant decisions, pharmacovigilance assessments or final MLR approval?
2. Override rule. Is there a written rule for when the team overrides or escalates an AI output, and does it name who signs?
3. Explainability. Can the manager explain an AI-assisted decision to quality assurance, an auditor or a regulator in two sentences?
4. Role redesign. Has at least one role or workflow in the team been rewritten around AI in the last six months, beyond simply being given a tool?
5. Coaching. Does the manager coach team members on AI use in one-to-ones, with protected time and a target, as part of line management alongside any central course?

A score of 8 to 10 marks a manager ready to lead the change. A score of 4 to 7 points to a targeted workshop on the gaps. A score of 0 to 3 means the team should start here, before it receives any further workforce-wide fluency module, because a fluency course delivered into an unchanged process produces exactly the access-without-use pattern Deloitte measured.

The check takes one function and one afternoon, which keeps it practical for a busy quarter. A low task-map score usually means the manager is rarely given those decisions to make, and that the central AI programme assumed someone else would.

Three moves put manager training ahead of workforce-wide fluency in the 2027 plan

For firms that budget on the calendar year, the 2027 enablement budget is being set this quarter. Each of the three readers who own a piece of it has one move to make.

• Heads of learning and development (L&D): pull last year’s AI training spend and split it three ways, across the whole workforce, technical specialists and people managers. The share that reached managers is the baseline to record for 2027.
• Function heads in commercial, medical or quality: run the five-question check with three of their managers this quarter and record the scores before any new course is bought.
• HR business partners: place one manager-layer workshop in the first quarter of the 2027 plan, ahead of the next workforce-wide fluency module, so that the people who redesign the work are ready before the next wave of users arrives.

This sequencing matches the case the Kainjoo Institute has made for purpose-built executive education in regulated industries⁶, where the scarce capability is people who combine compliance knowledge with the new tools.

Start where the work gets redesigned

Life sciences leaders have already done the visible part of AI adoption. They have bought the tools, written the strategy and tightened governance. Their own survey answers say the human side lags, with talent at 15% and falling, and their own sector report says managerial roles must move towards orchestrating human-AI teams. The enablement budget should follow that evidence and start with the managers who decide how each team’s work changes.

Kainjoo Institute designs tailored programmes for organisations, delivered as two- to three-day workshops, bootcamps or continuous learning. Teams planning 2027 can start with a workshop for the manager layer.

This article has been reviewed by Haider Alleg, Managing Partner at Kainjoo SA.

References

1. Deloitte AI Institute. State of AI in the Life Sciences & Health Care Industry: The untapped edge. March 2026. https://www.deloitte.com/content/dam/assets-shared/docs/about/2026/state-of-ai-lshc.pdf
2. Lyons, P., Konersmann, T., Jacobson, S., Kleyn, N., Rekhraj, K., and Gosalia, D. 2026 Life Sciences Outlook. Deloitte Insights, 9 December 2025. https://www.deloitte.com/us/en/insights/industry/health-care/life-sciences-and-health-care-industry-outlooks/2026-life-sciences-executive-outlook.html
3. Kainjoo. Case Managers Halve Discharge AI’s Error When the Bed Decision Arrives. kainjoo.life, 6 October 2026. https://kainjoo.life/discharge-prediction-ai-case-managers/
4. Kainjoo Institute. The AI Adoption Challenge: Go Beyond the Shiny Object in 2026. institute.kainjoo.com, 31 December 2025. https://institute.kainjoo.com/the-ai-adoption-challenge-go-beyond-the-shiny-object-in-2026/
5. Threlkeld, K. Employers Want AI-Native Talent, but the Workforce Isn’t There Yet. Here’s How to Close the Gap. Indeed Lead, 28 July 2026. https://www.indeed.com/lead/employers-want-ai-native-talent-but-the-workforce-isnt-there-yet-heres-how-to-close-the-gap
6. Kainjoo Institute. The Brand-Tech Skill Gap in Regulated Industries Is Structural, and Executive Education Is the Only Way to Close It Fast Enough. institute.kainjoo.com, 12 July 2026. https://institute.kainjoo.com/brand-tech-skill-gap-regulated-industries-executive-education/