Pressure, progress, and the AI question
Five years in, and the validation industry is anything but settled.
The 2026 State of Validation report — now in its fifth year, powered by Kneat Solutions — surveyed 614 validation professionals across six continents. The picture it captures is one of genuine tension: workload is at a five-year high, digital adoption is accelerating, and artificial intelligence (AI) is no longer a distant prospect. It has arrived. What the industry does next with these three converging forces will define the decade.
To mark the report’s launch, I moderated a panel discussion with four practitioners who each brought a different lens to the data — Durward Sutherland of IPS on workload, Ruairi McGarry of Kneat on the efficiency paradox, Michael Sweeney of PM Group on digital maturity, and Marcelo Nubile of Telstar on return on investment (ROI). Together, they brought the numbers to life. This article is a summary of what they shared.
Watch the panel discussion on demand now.
Why workload is back at a five-year high — and why it is not slowing down
80% of respondents report increased workload. That matches the 2022 peak, and it is not an anomaly. Durward Sutherland, Validation and Compliance Manager at IPS, broke down what is actually driving it.
The short answer: everything at once. Regulatory change is accelerating on both sides of the Atlantic. The EMA has published its three-year plan covering new guidance on advanced therapy medicinal products (ATMPs) for good manufacturing practice (GMP), new annexes including Annex 22 on AI, and continued pressure on data integrity. The US Food and Drug Administration (FDA) has signaled AI and machine learning guidance, along with eight additional guidance documents expected within the year. That alone represents a significant compliance workload — and it is just two regulators.
“That is just from two organizations, and it is quite a number of new documents that we will all have to digest and work with.”
Durward Sutherland, Validation and Compliance Manager, IPS
Beyond regulation, Durward pointed to a pipeline of new and increasingly complex products — biologics, monoclonal antibodies (MABs), antibody-drug conjugates (ADCs), and ATMPs — with radically different batch sizes and manufacturing approaches. In the UK alone, over 50 clinical trials are running on a single blood cancer. Every one of those trials generates data management requirements.
Talent is the third pressure point. Skilled validation professionals are difficult to find and retain. The skills the industry needs are changing as digitalization and AI mature, and data centers are competing hard for the same engineering talent. The transition from project delivery into ongoing operations remains chronically underestimated in terms of resource demand.
There is an irony here that Durward was direct about: some of the tools helping organizations get more efficient — digital validation platforms, AI integrations, new enterprise software — also add to workload in the short term. The learning curve is real, and it has to be planned for.
The efficiency paradox: more work, but more efficient
Here is the figure that stops people: 67% of respondents say they are more efficient than a year ago. That is the third consecutive year of that finding. Set it alongside record workload, and the question becomes obvious — how is that possible?
Ruairi McGarry, Senior Product Owner at Kneat, had a clear answer: digital adoption has fundamentally changed the capacity equation. The industry is doing more, but it is doing it with tools that did not exist a few years ago.
“Seven years out of life sciences directly — and in that period, the changes have been absolutely enormous. There was very little digital adoption when I was working there. It was all Microsoft Word® documents. The tools have come on leaps and bounds.”
Ruairi McGarry, Snr. Product Owner, Kneat
Ruairi pointed to a significant structural shift in the top reported challenges. In 2022 and 2023, the top concerns were shortage of human resources, efficiency, and lack of resources. By 2024, 2025, and 2026, those three had dropped out entirely. The new top challenges — compliance burden, audit readiness, and data integrity — reflect an industry that has largely solved the resource-constraint problem through technology and is now focused on higher-order compliance and quality challenges.
This is not a trivial observation. It means that digital tools are genuinely enabling people — not just in theory, but in practice, at scale, and across years. The efficiency gains are real. The question now is whether organizations are using that efficiency dividend to get ahead of the compliance complexity coming at them, or simply treading water.
Digital maturity: from experimentation to industrialization
The report records a major jump in full digital validation adoption — from 16% to 30% year over year. That is a near-doubling in a single year. Michael Sweeney, Electronic Validation Lead at PM Group, read it as a signal of fundamental change.
“It shows the industry is moving beyond experimentation into scale. We are transitioning from pilot phases to adoption at scale. Digital validation is becoming embedded in standard ways of working.”
Michael Sweeney, CQV e-Validation Lead, PM Group
Michael identified three forces behind the jump. First, confidence: early adopters have now proven that digital validation tools withstand real regulatory inspections, removing a significant barrier for organizations sitting on the fence. Second, business pressure: timeline constraints, resource costs, and the need for reusable validation assets are driving practical decisions. Third, scale: companies are no longer treating digital validation as an innovation initiative — it is becoming core infrastructure.
But 70% of organizations still have not fully adopted digital validation. Michael was clear-eyed about why: change management is hard, skill gaps are real, process redesign takes time, and data governance questions do not resolve themselves. Most organizations are operating in a hybrid paper-and-digital state, and that is not automatically a failure — it is a transition.
13% of organizations report being fully digital across all record types. Logbooks, batch records, and validation protocols remain predominantly paper-based. Durward reinforced why: legacy manufacturing equipment was not designed to capture digital data automatically, and replacing working systems for the sake of digital records is rarely economical. The shift happens as systems are upgraded — not as a big-bang transformation.
Michael’s forward-looking point is worth holding onto. The question is no longer just whether organizations are digital — it is whether their data is working for them. Isolated digital tools are a first step. End-to-end integration, where data flows across systems and drives real-time insight, is where the industry is heading.
The ROI story: expectations met, and increasingly exceeded
74% of respondents who have had sufficient time to assess their investment in digital validation say it met or exceeded expectations. 30% say it significantly exceeded them — nearly double the prior year’s figure. Marcelo Nubile, Consultancy and IT Manager at Telstar Brazil, described exactly how this plays out on the ground.
Marcelo worked with a client operating multiple sites across Latin America who had implemented Kneat. Their initial ROI expectations were conservative — around a 30–40% reduction in validation execution time. What they found, after the learning curve of the first year, was significantly higher returns by year two.
“In the beginning, you have a learning curve. You spend time on training, on implementation — so you do not see a lot of return in the first year. But in the second year, the software is more mature, you are more confident. That is why we are seeing this great return on investment now.”
Marcelo Nubile, Consultancy & IT Manager, Telstar Brazil
This is an important qualification for organizations evaluating digital validation investment. Year-one ROI is rarely the right measurement. The system is maturing, teams are building confidence, and workflows are being redesigned. The compounding value emerges from year two onward.
The report also captured a notable shift in where organizations are seeing value. Better traceability has moved to the top of reported benefits, displacing data integrity for the first time. Marcelo explained why this resonates in his market: traceability reduces the time spent locating documents and evidence, which directly reduces the cost of executing validation activities. In an environment where companies are under sustained cost pressure, that is a tangible, boardroom-level result.
AI has arrived — and accountability is the real question
39% of respondents are using or evaluating AI for Good Practice (GxP) validation — 22% are actively deploying it. “Not considering AI” collapsed from 46% last year to 19%. The fear factor has not disappeared, but the default position has shifted. AI is no longer something the industry is watching from a distance.
Michael described where use cases are actually emerging. The four areas PM Group is seeing in client work are test script generation and optimization, requirements traceability matrix automation, deviation and test failure classification, and document review and compliance checks before human review begins. He was clear about the value and the limits.
“AI adoption is real, but what we are seeing in the field is still targeted and pragmatic. It is not yet transformational. These are focused, low-risk use cases — but they still have tremendous value.”
Michael Sweeney, CQV e-Validation Lead, PM Group
The top AI concerns among respondents are accuracy (58%), data security (50%), and hallucinations (45%). Ruairi addressed these directly and without minimizing them. These are not abstract worries — they map precisely onto what GxP compliance demands every day. His answer was not to wait for the concerns to resolve, but to build governance around them.
“Companies are going to have to ensure they have the correct securities in place. If you want to keep your data secure, you need to make sure the AI models are not training on your data if you do not want them to be. You have to work with your suppliers on this.”
Ruairi McGarry, Snr. Product Owner, Kneat
Ruairi also referenced the first FDA warning letter related to AI overreliance in a medical device company — issued in April 2026 — as a concrete signal that regulators are watching and acting. The subject matter expert (SME) must remain at the center of decision-making. AI is assistive. It is not a replacement for experienced judgment, and no organization should treat AI outputs as approved without scrutiny.
On regulatory readiness: 36% of respondents describe themselves as only slightly familiar with, or not familiar at all with, emerging AI regulatory guidance. Annex 22 from the EMA — focused on AI in GMP systems — is out for review but may not be finalized for another year. FDA guidance is in development. The industry is, in a real sense, deploying AI ahead of the formal regulatory framework that will govern it.
Durward’s perspective on this was grounded: the industry has navigated exactly this situation before. When Annex 11 was updated, practitioners debated successive drafts and adapted their approaches before the final version was published. The same will happen with AI guidance. The industry will track the evolving framework, share intelligence across organizations, and adapt. The collective experience in this space is not nothing — it is a real asset.
78% of respondents believe AI will be standard in validation by 2030. That is four years away. The gap between that expectation and the current 36% regulatory familiarity rate is where the work needs to happen. Closing it requires investment in training, in understanding the foundational guidelines already in place — ICH Q7, ICH Q9, Good Automated Manufacturing Practice (GAMP), computer system validation (CSV) — and a disciplined approach to building on those fundamentals before reaching for new technology. Marcelo put it simply: standardization of how AI is used is not a future problem. It needs to start now.
What the fifth year tells us
Five years of this research has built something the validation industry has not had before: a longitudinal dataset that tracks real change over time. The trajectory is clear. Digital adoption has moved from novelty to standard practice for a significant portion of the industry. ROI is being realized and increasingly exceeded. AI is no longer aspirational — it is operational, in specific, targeted ways, inside real validated environments.
The pressure remains. Workload is high, regulatory expectations are rising, and the pace of change in pharmaceutical development is unrelenting. But the tools are better, the data is clearer, and the community is more connected than it has ever been. The industry is not in crisis — it is in a period of genuine, sometimes difficult, transformation. That is different.
The full 2026 State of Validation report is available now at stateofvalidation.com.
