UNLEASH 2025: The AI HR Moves Leaders Copy in 2026

A lot of HR teams spent 2025 talking about AI. The companies that actually benefited from it did something less flashy: they treated AI as workforce infrastructure—something you design, govern, measure, and continuously improve.

UNLEASH’s most-read stories from 2025 make that pattern hard to ignore. Candidate experience showed up right next to billion-dollar upskilling bets. HR and IT org charts got rewritten. People analytics platforms moved from “nice to have” to “how we run the business.” And in the background, the DEI conversation became more politically charged, forcing leaders to separate values from optics.

This post pulls the practical lessons from those UNLEASH highlights and translates them into a 2026 action plan—especially for leaders in AI in supply chain & procurement, where workforce capacity, skills availability, and operational execution determine whether AI initiatives deliver real ROI.

The 2025 pattern: AI in HR is turning into operating discipline

Answer first: The big shift in 2025 was that AI in HR stopped being a tool conversation and became an operating model conversation.

UNLEASH’s top stories weren’t dominated by “cool demos.” They were dominated by questions like:

• How do we design a candidate experience that competes with consumer-grade apps?
• How do we upskill hundreds of thousands of people without turning learning into a checkbox?
• Who owns the flow of work when AI touches every workflow?
• How do we prove value (and reduce risk) when AI decisions affect humans?

If you run supply chain or procurement transformation, this should feel familiar. AI demand forecasting fails when planners don’t trust it. Supplier risk models fail when nobody owns the process after the alert triggers. The same dynamic is now playing out in HR: adoption is less about model quality and more about workflow ownership, change management, and governance.

Snippet-worthy truth: “AI in HR succeeds when it’s treated like a business system, not a productivity hack.”

Candidate experience is now a competitive advantage (not a TA slogan)

Answer first: The Medtronic story shows a core 2025 lesson: candidate experience is where AI automation either builds trust—or destroys it.

Medtronic’s recruiting challenge (low brand awareness despite massive scale) is basically the talent version of selling a great product no one recognizes. Their approach points to a broader principle: automation should make the process feel more human, not more robotic.

What “good automation” looks like in hiring

AI in talent acquisition works when it reduces uncertainty and waiting.

Practical moves you can copy:

• Use AI to shorten the “silent gap.” Automate status updates, interview scheduling, and next-step clarity.
• Personalize the journey without creeping people out. Tailor content by role family and location, not by overly specific behavioral predictions.
• Design for closure. A fast “no” with dignity beats a slow “maybe” with no information.

Why this matters to supply chain & procurement

If you’re building AI-driven supply chain planning, you already know the adoption rule: people trust systems that are predictable.

Hiring works the same way. And it’s directly linked to operational outcomes:

• Faster fills reduce overtime and expedite fees.
• Better quality of hire reduces early attrition (which is expensive in plant and warehouse roles).
• Stronger candidate communications reduce offer drop-off—critical when specialized procurement talent is scarce.

If your 2026 workforce plan depends on hiring AI-literate planners, category managers, or data engineers, candidate experience isn’t branding. It’s capacity planning.

Upskilling is the real “AI investment”—and EY proved it with $1.4B

Answer first: EY’s $1.4B AI investment story is a reminder that your biggest AI cost isn’t software—it’s workforce readiness.

Buying tools is easy. Changing how 10,000 (or 400,000) people work is the hard part.

Here’s what I’ve found in real implementations: the upskilling programs that work treat learning like a product. They have a target persona, a clear outcome, usage metrics, and iteration cycles.

A 2026 upskilling blueprint you can run without “training theater”

If you want AI literacy that shows up in execution (not just course completions), build three layers:

1. Baseline AI literacy for everyone

   • What GenAI is and isn’t
   • Data privacy, IP, and prompt hygiene
   • How to verify outputs and cite sources internally

2. Role-based AI capability for critical job families

   • Recruiters: structured interviewing + AI-assisted sourcing governance
   • HRBPs: policy interpretation and employee comms review workflows
   • Supply chain planners: scenario planning, demand sensing interpretation
   • Procurement: supplier risk triage, contract review checkpoints

3. “Builders” pathway for power users

   • Lightweight automation (workflow tools, internal copilots)
   • Model evaluation basics (bias checks, drift awareness)
   • Process design: where humans must stay in the loop

The metric that beats course completion

Track time-to-proficiency instead:

• How long until a planner can run a scenario plan with AI and explain assumptions?
• How long until a recruiter can justify why a candidate moved forward without referencing “the AI said so”?

Completion rates are vanity. Proficiency time is operational.

HR + IT is merging because “flow of work” is the new battleground

Answer first: Moderna’s HR/IT merger is a signal that companies are reorganizing around workflow ownership—because AI touches everything.

When HR runs policy, IT runs tools, and each blames the other for adoption, AI programs stall. Moderna’s move reflects what many leaders are quietly realizing:

• AI copilots live inside productivity suites.
• People data lives across HRIS, identity, security, and collaboration platforms.
• Governance requires both human policy and technical enforcement.

What to copy (even if you don’t merge departments)

Most companies don’t need a full org merger. They do need shared accountability.

A workable model:

• HR owns: use-case prioritization for people processes, policy, experience design
• IT owns: identity, access, security controls, vendor risk, integrations
• Jointly own: workflow design, monitoring, incident response, change management

This is especially relevant in AI in supply chain & procurement, where AI tools often sit across ERP, planning suites, supplier portals, and collaboration tools. Ownership gaps are where risk and waste hide.

Snippet-worthy truth: “AI isn’t a feature you deploy. It’s a workflow you run.”

People analytics is getting productized (Spotify’s move is a tell)

Answer first: Spotify’s people analytics platform story shows the next phase: analytics becomes an internal product with external-grade expectations.

When leaders rely on workforce metrics for decisions (headcount, skills, retention, productivity), the analytics layer can’t be a spreadsheet side hustle. It needs:

• A defined data model
• Governance on definitions (what counts as attrition? internal mobility?)
• Secure access patterns
• Repeatable dashboards and insights

How to connect people analytics to AI ROI

If you’re trying to justify AI spending, you need metrics that finance respects.

A simple chain:

• Adoption metrics (who uses the tool, how often)
• Process metrics (cycle time, error rate, rework)
• Business metrics (cost per hire, retention, fill rate, overtime)

For supply chain & procurement, extend that chain into operational KPIs:

• Forecast accuracy improvements tied to planner adoption
• Reduction in expedite shipments tied to better capacity planning
• Supplier disruption response time tied to risk workflows

The point: workforce analytics and operational analytics are converging. Leaders who keep them separated will struggle to prove AI value.

The uncomfortable 2025 question: Is AI weakening thinking?

Answer first: The “AI reduces cognitive and creative skills” debate matters because HR is now responsible for preventing skill atrophy—not just building new skills.

When AI drafts every email, summarizes every meeting, and proposes every first idea, people can lose reps in:

• Structured reasoning
• Writing clarity
• Problem framing
• Creative synthesis

This isn’t a reason to ban tools. It’s a reason to design usage norms.

A practical “anti-atrophy” policy that won’t get ignored

Try three rules that managers can enforce without policing:

• Draft-first rule for critical thinking tasks: bring your own first pass, then use AI to refine.
• Verification rule for decisions: AI output must be paired with a human-stated rationale and a data source.
• No-AI zones: certain artifacts (performance reviews, disciplinary notes, final supplier decisions) require documented human judgment, with AI only assisting formatting or summarization.

For procurement teams, this is especially important in negotiations, supplier relationship management, and risk calls—areas where judgment and context beat pattern matching.

DEI rollbacks and AI governance: the same trust problem

Answer first: DEI rollbacks and AI governance collide on trust: employees watch whether leadership decisions match stated values.

In 2025, UNLEASH highlighted companies rolling back DEI commitments amid political pressure. HR leaders can’t treat this as “separate from AI.” Here’s why:

• AI systems can amplify inequities if training data and processes aren’t governed.
• If employees suspect AI is being used to screen them unfairly, trust collapses.
• Trust collapses adoption—then your AI investment becomes shelfware.

What good governance looks like in 2026

Keep it simple and enforceable:

• Declare where AI is used in people decisions (recruiting, internal mobility, performance, scheduling).
• Document human checkpoints (who can override, who reviews outcomes).
• Audit outcomes (selection rates, promotion rates, performance distribution shifts).

This is the same governance mindset supply chain teams use for supplier compliance: define standards, monitor outcomes, and escalate exceptions.

A 30-day plan for HR + operations leaders heading into 2026

Answer first: You don’t need a 12-month AI roadmap to make progress—you need four decisions in the next 30 days.

1. Pick one workflow where AI reduces cycle time

   • Example: frontline hiring scheduling; procurement intake triage; planner scenario generation

2. Define success with three numbers

   • One adoption metric, one process metric, one business metric

3. Name a single workflow owner

   • Not a steering committee. One accountable leader.

4. Write the “human-in-the-loop” rule in plain language

   • What AI can do, what it can’t do, and what must be reviewed by a human

If you do only this, you’ll enter 2026 with momentum—and a foundation you can scale.

Where this fits in AI in Supply Chain & Procurement

AI in supply chain & procurement lives or dies on execution: planners must trust forecasts, buyers must trust risk signals, and operations must trust the workflow.

UNLEASH’s 2025 HR highlights are a reminder that the workforce side is now inseparable from the operational side. The companies winning with AI aren’t only improving algorithms—they’re improving how people adopt, govern, and benefit from them.

If you’re planning your 2026 initiatives, ask yourself one forward-looking question: Which workflow will you redesign first so AI becomes part of how work runs—rather than another tool people ignore?