| Last Week in HR AI: Issue #20 | Week of November 9, 2026 |
36% of HR teams using AI see the same results as non-users
They rolled it out ahead of their process and data. Sapient's 29th survey covers 6,038 organizations.
|
1 ai.wharton.upenn.edu2 Sapient Insights3 i4cp4 maccus.fr
Ikona's Take on the past week (November 9, 2026) in HR + AI
|
IO
|
Ian O'Keefe
Co-founder & CEO, Ikona Analytics
|
The pattern this week is plain: AI pays off where the organization redesigns the work around it, and stalls where it is handed out ahead of process, data, and ownership. Sapient Insights Group's 29th HR Systems Survey puts a number on the stall: 36% of AI users adopted ahead of their processes and data, and their outcomes look like non-users'. License counts are not adoption, and they are not a business case.
S&P Global went the other way. Girish Ganesan started with five cross-functional workflows, grew to nine covering about 70% of employees, and built a leadership academy for 7,000 managers. Japan's Gennai shows the alternative: weekly users report saving roughly 110 minutes a week, while about 31% rarely or never use it. Access without redesign builds a productivity gap inside your own workforce. Bocconi's Copilot study names the next constraint: when answers are cheap, validating them is the work.
Which leaves the uncomfortable question: who owns the redesign? Vendors now ship agents that act inside HR systems, and the HR Tech 2026 accounts disagree on who answers for them. Forward deployed engineer postings jumped from under 200 in 2023 to over 7,300 in the first nine months of 2026, because someone has to wire AI into real work. If you cannot name the approver, the data, and the shutoff owner for each workflow, that someone is not you yet. I'd like to hear where your organization sits. Reply with your take, or reach out and we can dig in together.
| The six to read | The stories that matter for the Office of HR — with our take on each. |
In its fourth annual survey, Wharton Human-AI Research polled about 1,400 employees at large U.S. companies on their use of artificial intelligence (AI). Use is widespread, but only 7% of organizations combine strong return on investment (ROI), significant workflow change, and alignment. Companies that track growth metrics, and those with a dedicated chief AI officer, report better returns. Managers report more alignment and better training than frontline employees do, and accountability for errors made by AI agents is still unclear.
For its 29th annual survey of human resources (HR) systems, Sapient Insights Group collected more than 10,000 responses from 6,038 organizations across 23 HR technology categories. It reports that 55% of organizations use AI in HR, while 36% of AI users put it in place before their processes and data were ready and report outcomes close to those of non-users. Executives now lead 31% of system selections, and most change management practices declined. The full report is priced at $1,800.
i4cp profiles Girish Ganesan, chief people officer at S&P Global, on how the 40,000-person company is reorganizing work around AI. The effort began with five cross-functional workflows and has expanded to nine, which together cover about 70% of employees. S&P Global also set up three HR capability centers and a leadership academy for 7,000 managers. Ganesan organizes the work around four questions: where AI adds value, where human judgment remains, how to use freed capacity, and how to govern agents as part of the workforce.
Also covered by: accenture.com
|
From Ikona
AI stalls when the work underneath it is undocumented
We diagnose how work and decisions actually flow through your HR function and capture the judgment that lives in people's heads. Then we turn it into a structured knowledge asset your AI workflows can stand on. Talk to us about yours → |
Japan's Digital Agency surveyed 6,439 staff about Gennai, a generative AI tool available across government to about 180,000 employees. The 47.8% of respondents who use it weekly report saving roughly 110 minutes a week on delegated tasks, a figure estimated from self-reported ranges. About 31% say they rarely or never use it. The agency is now adding agent tools and an app for drafting parliamentary answers, and it has warned of a productivity gap between users and non-users.
Researchers at Bocconi followed behavioral data for 12,040 employees at a consumer products multinational for 13 months during a staged Microsoft Copilot rollout. Copilot users spent more time collaborating, worked with a wider set of colleagues, and had less clustered networks. They also relied more heavily on colleagues they already knew well. The authors conclude that AI moves the bottleneck from producing answers to validating them, reshaping how people depend on one another rather than replacing that dependence.
Four write-ups from HR Tech 2026 agree that vendors are now shipping agents that take actions inside HR systems. They disagree on where accountability sits: with the platform, the manager, the labor cost of review, or a named owner. Oracle's release keeps employment decisions with the employer. The piece suggests buyers check four things: a demo of the agent log, a named approver for each workflow, an estimate of review hours, and an assigned shutoff owner.
Also covered by: dhrmap.com, futurumgroup.com
Argues that most people analytics functions have strong system data but no layer describing how work actually gets done, and that this missing tacit knowledge is why generative AI projects in HR stall. Owning that context layer, it says, is the next strategic claim for analytics leaders.
Only 7% of organizations in Wharton's survey show strong returns, real workflow change, and alignment at the same time, and S&P Global's path started with five workflows, not a license count. So on Monday, pick one workflow where usage has stalled and run the four checks from the HR Tech 2026 piece against it: see the agent log, name the approver, estimate the review hours that Bocconi's validation bottleneck will actually cost, and write down who can shut it off. If filling in those four lines takes longer than the vendor demo did, you have found exactly where your redesign starts.
If your team is wrestling with workforce intelligence, AI readiness, or a transformation that has to land, we'd love to help. Reach out and tell us what you're working on.
Get in touch →Last Week in HR AI
Practitioner perspectives and diagnostic frameworks, delivered the week they publish. Be the first to know when new research drops.
No spam, ever. Unsubscribe anytime.