Five research and vendor releases published on 19 November describe the same pressure from different sides: companies are deploying AI faster than they are deciding who shapes the work around it. The Adecco Group’s survey of 40,500 workers found that 96% of workers who were involved in redesigning their work are willing to adapt to change, against 29% of those who were not. Four other releases that day show where that involvement is being built, and where it is still missing.

The Adecco numbers: adoption is high, involvement is uneven

The Adecco Group (SIX: ADEN) published the seventh edition of its annual Global Workforce of the Future report on 19 November. The survey covered 40,500 workers across 36 countries and 21 industries, fielded online in local languages between June and July 2026. Nearly three-quarters of respondents, 74%, use AI agents at work, up from 64% in 2025.

The report’s central finding is about integration. Workers who say AI is well integrated into operating workflows report saving 20% more time per day than workers with lower integration. They are also more likely to reinvest the saved time in higher-value work, and seven times less likely to feel overwhelmed by the need to use AI tools effectively. Adecco draws the conclusion that productivity gains depend on how AI is embedded in the work, more than on who has access to it.

Advertisement

HRTech Your brand belongs here. Reach the decision-makers who read HRTech every day. Premium placements across the site and newsletter. Advertise with us

The involvement figures are the sharpest in the release. Among workers whose employers involve them in redesigning work in response to AI, 96% say they are willing to adapt to changing environments. Among workers who are not involved, the share is 29%. Communication shows a similar split: 92% of workers whose leaders explain how AI will change their role are comfortable with AI-related change, compared with 15% where that explanation is missing.

Denis Machuel, CEO of the Adecco Group, said in the release: “When workers help redesign how work gets done, and receive the training and clarity to adapt, AI creates more value and change becomes easier to navigate.”

Two more findings bear on HR budgets. Two-thirds of workers, 66%, say AI has changed the skills their jobs require, yet participation in AI training fell from 77% in 2025 to 67% in 2026. Among AI users, 93% of those who completed training say AI has made them more productive, compared with 47% of those who have not trained. Adecco calls the weak link a leadership bottleneck: only 19% of workers say leaders are empathetic to the effects of change, 21% feel concerns about new technologies are taken seriously, and 39% feel supported by their manager to adapt.

The sample is workers, so every figure is a self-report. The release does not say how “well integrated” was defined in the questionnaire, and it does not tie any figure to a particular employer or tool. The causal language in the release is Adecco’s interpretation of a single survey.

Who sits in the room: RedThread on L&D

RedThread Research, a human capital research and advisory firm, released a study the same day titled L&D and AI: How L&D changes because AI exists in the organization. The study draws on in-depth interviews with learning and development leaders, research roundtables and a review of 50 sources, and it catalogs 44 “moves” organized across five dimensions: Direction and Vision, Development Experience, Operations and Governance, Tech and Data, and Stakeholder Alignment.

RedThread describes the organizational setting plainly. The C-suite in most organizations wants everyone to use AI, yet no one has defined what that means or what “AI fluency” means. AI strategy usually sits with the CIO, a transformation office or a Chief AI Officer, and when L&D is pulled in, it is often with a vague remit to “train everyone on AI.”

That context lines up with Adecco’s training numbers. A learning function handed that remit has no say in how the work itself changes, and Adecco’s drop in training participation is the cost. RedThread’s leaders respond by setting direction early and getting into the forums where AI decisions are made. Twenty-two of the 44 moves, half, need little or nothing in place to start. Five of those change what L&D does, including inviting itself into AI planning and designing for judgment, which RedThread defines as deliberately building the judgment AI can erode.

Dani Johnson, Co-Founder and Principal Analyst at RedThread Research, said the leaders it interviewed are not stopping at tool questions: “They’re asking bigger, more strategic questions, like what L&D needs to become now that AI simply exists in their organizations.”

RedThread offers two examples. A talent management leader gave managers an AI prompt pack built around a 30/60/90-day framework to use directly with each new hire. A global head of learning found that leaders across her company were already building their own courses with AI and that guidelines alone would not shape the results, so her team is building an agent that handles end-to-end learning design for people outside L&D, with outcome definition, measurement and design practices built in. The full report is for RedThread members, so this piece relies on the public release.

Decision rights: McLean and Company on L&D governance

McLean and Company, an HR research and advisory firm and a division of Infotech, published its Guide to Establishing Learning and Development Governance on 19 November. The guide organizes governance around three building blocks, decisions (what requires oversight), people (who is accountable) and processes (how governance runs day to day), and walks HR leaders through four areas: identifying the decisions that need governance, defining a decision structure, establishing governance processes and sustaining the framework.

McLean’s guide asks leaders to decide whether learning decisions are centralized, decentralized or shared, and to assign roles for input, approval and execution. It says less complex organizations may be well served by governance led directly by HR or L&D, while complex organizations may need a cross-functional council.

Leann Schneider, director of HR Research and Advisory Services at McLean and Company, said in the release: “The goal is to make ownership visible, involve the right people, and create practical routes from a learning request to a well-supported decision.”

The release does not mention AI. It applies here because RedThread’s example of leaders building their own AI-made courses is a decision-rights problem: someone has to say which courses count and who reviews them. McLean’s language gives HR a way to answer in writing.

Involvement in practice: Koch and HappyRobot

The fourth release shows what involvement can look like in a deployment. Koch announced a partnership with HappyRobot, a company that builds AI agents for enterprise operations, to test how agents can support employees and candidates. The release came from Koch Disruptive Technologies, which the release lists among HappyRobot’s investors; HappyRobot has raised approximately $200 million in total.

The project began with a two-day workshop in Wichita a few months ago. According to the release, HR professionals, business leaders and the HappyRobot team worked side by side and started with the problems. They mapped where employees and candidates wait for answers and where HR professionals repeat the same tasks. By the end of the second day the group had identified a handful of use cases and built two working prototypes.

Two applications are in testing. In recruiting, an AI agent conducts initial candidate prescreening calls, confirms key qualifications and captures information for recruiters. In the HR service center, Koch and HappyRobot are exploring how agents can help employees with common HR questions, including payroll and tax navigation inquiries.

Jeff Wendling, recruiting innovation lead at Koch, said of the workshop: “Within a couple of days, we had a working solution that was already 75% to 80% of the way toward what we were trying to accomplish.” Cara Chennault-Reid, vice president of human resources at Koch, described the program as experimentation: “We are experimenting, learning and finding new ways of working to position us for the future.”

Newsletter

Get the week's best tech coverage.

Free. Read by thousands of HR, tech, and business leaders.

The release is a partnership announcement from one side of the deal, and it reports no results. It gives no screening volumes, no accuracy or candidate-experience measures and no figures on HR service-center deflection. What it does document is process: HR staff who do the work helped decide what to build before anything was deployed, which is the practice Adecco’s involvement numbers point toward.

Embedded AI and the training gap

A fifth release the same day shows the vendor side of integration. Würk, a human capital management company serving cannabis, beverage alcohol and other highly regulated industries, launched Bryte AI, a suite of embedded AI capabilities and agents that works inside its payroll, HR, time and scheduling platform. Bryte Embedded AI automates routine tasks such as creating job descriptions, scheduling applicant interviews and drafting performance review comments, and Bryte AI Assistant answers HR, payroll, time and policy questions in plain language. Würk says Bryte AI is available to all of its customers.

Dan Obendorfer, VP of Product and Technology at Würk, said in the release: “Our clients operate in some of the most heavily regulated industries in the country, so every capability we ship has to reduce risk and save time at the same time.”

Embedding AI in the HR system matches what Adecco describes as integration into workflows. Adecco’s training data adds a condition: trained users report far higher productivity. Buyers evaluating embedded assistants can ask a vendor which tasks run in the background, which need a person’s approval, and what the customer’s own staff must learn to supervise them.

What it means for the HR leader

The five releases come from a worker survey, a research firm, an advisory firm and two vendor announcements, so they do not prove one another. Read together they point to a sequence of decisions.

First, decide who owns the AI decisions that touch work. RedThread’s finding that AI strategy usually sits with the CIO, a transformation office or a Chief AI Officer means HR often learns about changes to work after they are scoped. McLean’s decisions, people and processes structure can be applied to AI-related requests: which ones need approval, who gives it and how a request moves from intake to a decision.

Second, involve the people whose work changes before the design is fixed. The Koch workshop shows one way to do it, with the people doing the work in the room on day one. Adecco’s 96% and 29% figures are the argument for the time that takes.

Third, check whether guidance reaches employees. Our earlier coverage of TriNet’s survey found 90% of employers saying they gave AI guidance while 66% of employees said they received it. Adecco’s 92% comfort where leaders explain role changes, against 15% where they do not, makes delivery of that explanation worth measuring. The LSE and Protiviti study we covered, in which 48% worried about being judged for not using AI enough, describes the pressure on the employee side of the same gap.

Fourth, put the training budget where the skills are changing. Adecco’s 66% who say their skills requirements changed, and the drop in training participation from 77% to 67%, are the plainest numbers for a budget conversation. The 93% against 47% productivity split between trained and untrained users is a self-report that an HR team can test with the same question in its own pulse survey.

Fifth, give the people who manage AI work a defined role. When Lattice unveiled Pace, it assigned each AI agent a human manager and a goal; Adecco found only 39% of workers feel supported by their manager to adapt, so the managers need time and guidance for the role.

In our reading, the common thread is that none of the five releases reports a result from letting workers shape the design that a skeptical buyer could verify. Adecco reports attitudes, RedThread reports leader behavior, McLean offers a framework, Koch reports prototypes and Würk reports a launch. HR leaders who want evidence will need to collect it themselves, by recording who was involved in each redesign and comparing adoption and satisfaction against teams where nobody was asked.

What to watch

The Adecco report’s full tables are the first place to check how “involved in redesign” and “well integrated” were defined. Koch and HappyRobot say testing continues and other functions are under consideration, so usage figures would turn their process story into data. Until then the evidence is a large survey and a set of early practices, and the practical move is to put decision rights, training and worker involvement in place before the next AI rollout reaches a team.

Source: The Adecco Group, “The Adecco Group global study: AI delivers more value when people are involved in work redesign,” 19 November 2026. Also drawn from releases by RedThread Research, McLean and Company, Koch Disruptive Technologies and Würk, all dated 19 November 2026.