The Talent Algorithm — Issue No. 01: The quietest signal of AI's impact on work isn't a layoff

Two months ago, JPMorgan Chase held its February 24 investor meeting. Jamie Dimon, in an exchange with analysts, said something most corporate CEOs have been careful not to say out loud. “We already have huge redeployment plans for our own people,” he told the room. “We have displaced people from AI — and we offer them other jobs.”

That last clause is the one to sit with. Not the redeployment plan. Not the technology spend. The admission that AI is already displacing workers at the largest bank in the United States, and the posture JPMorgan has taken in response: move them, don't cut them.

Beneath JPMorgan's roughly steady total headcount of 318,500, the composition is shifting. CFO Jeremy Barnum, presenting at the same meeting, described the pattern: Operations roles are down roughly 4%. Support functions are down roughly 2%. Revenue-generating and client-facing roles are up roughly 4%, offsetting the reductions almost exactly. AIHR's 2026 HR trends report finds that 89% of HR functions have already restructured or plan to in the next two years. JPMorgan is showing the field what one version of that restructure actually looks like — not on a consulting slide, but in the live composition of a 318,500-person workforce.

I think this is the signal that matters for 2026. Not the cost savings. Not the technology stack. The redeployment posture itself. For a decade, the default HR response to automation has been attrition and severance — smooth the exit, minimize the legal exposure, report the savings. JPMorgan is doing something structurally different, and doing it publicly. The CEO is naming displacement by its real name, and the company is putting itself on the hook to find the displaced a second act inside the firm. That is not a PR move. That is an operating model choice with real cost implications and a real bet: that retained institutional knowledge is worth more than the short-term margin from cutting.

Here is what I find genuinely exciting about this moment, and I want to be direct about it because the field has spent a lot of time being worried. When AI handles sourcing, screening, scheduling, transaction processing, and first-pass analysis, the work that remains is the work most HR professionals say they went into this field to do. Coaching leaders through change. Designing redeployment paths before the automation ships, not after. Building the skill infrastructure that makes a second act possible. Thinking in years rather than quarters.

The practical implication for CHROs and VPs reading this: the question is no longer whether AI will displace work inside your organization. Dimon just said the quiet part out loud on behalf of the whole field. The question is what your displacement response looks like. Do you have a redeployment infrastructure — skills maps, internal mobility pathways, retraining budgets, a named owner — ready to absorb the first wave? Or will your first agentic AI rollout arrive at an HR function that is architected for hiring and exiting, not for moving people sideways at scale?

What I want The Talent Algorithm to be, starting with this issue, is a resource for that redesign. Not a feed of warnings about what could go wrong — the HR press is saturated with those. A weekly brief on how to actually do the work: the models, the frameworks, the prompts, and the decisions that turn AI from a threat you manage into an asset you run. The ground has moved. The question is no longer whether to redesign the function. It's what to redesign it around. Let's get into that.

· · ·

In a late-March 2026 memo obtained by Business Insider, Microsoft Chief People Officer Amy Coleman laid out a reorganization of the HR function supporting the company's 220,000-person workforce, including a new unit called Workforce Acceleration. Per the memo, the new team covers “skilling, redeployment, workforce planning, and the emerging human-agent collaboration.” Trade coverage including Reworked has noted the function has no direct precedent at Microsoft or its peers. If JPMorgan is the first institutional-scale example of the redeployment posture in action, Microsoft is the first visible example of the HR structure being redesigned to support it. Two different industries, same directional move.

Source: Business Insider memo coverage; HR Grapevine; Reworked

AIHR's 2026 HR trends report argues the traditional HR structure — separate columns for Talent Acquisition, L&D, Total Rewards, and Performance — is collapsing into more integrated models as AI platforms erode the boundaries between those functions. The report names Workday, SAP Joule, and Microsoft Copilot as examples of systems now pulling data and workflows across the full employee lifecycle. Strategic implication: if your org chart still shows four clean pillars, it is probably already out of sync with how work actually flows through your HR tech stack.

Source: AIHR — 11 HR Trends for 2026

SHRM's State of AI in HR 2026 report, based on responses from 1,908 HR professionals, finds 73% of talent acquisition leaders ranking critical thinking as their top hiring priority — with AI technical skills ranked noticeably lower on the same list. The shift is not away from AI. It is toward the judgment that determines whether AI is used well. Worth keeping on hand the next time you rewrite a job description for your own team.

Source: SHRM — State of AI in HR 2026

Two regulatory signals belong in your risk register. First: core enforcement of the EU AI Act's obligations for high-risk AI systems begins August 2, 2026, and under Annex III of the Act, most HR AI — including systems used for recruitment, selection, promotion, and performance evaluation — is classified as high-risk, triggering requirements around human oversight, bias testing, and documentation. The Commission's Digital Omnibus proposal could shift some deadlines, but the direction of travel is set. Second: in the US, the Mobley v. Workday case was granted conditional class certification in March 2026 — the first major class action on AI hiring bias to clear that bar, and a clear signal that both deployers and vendors now carry real exposure.

Source: EU AI Act (Annex III); SHRM and HR Dive coverage of Mobley v. Workday

· · ·

A new series I'm kicking off this week. Over the next four issues, I'll give you a framework, a process playbook, a working prompt, and a self-assessment — in that order — to help you actually move on AI-era HR, not just read about it. We start with redeployment readiness, because the JPMorgan signal is only actionable if your HR function can absorb what it implies.

Dimon's comment describes an outcome. It doesn't describe the infrastructure that makes the outcome possible. Here are the four capabilities every HR function needs to own clearly before its first agentic AI rollout ships, regardless of company size:

1. Skills visibility.

What skills does the organization actually have today, and what will it need in 12–24 months? This is no longer an L&D question. It is a workforce planning question that L&D executes. You need a live view of both sides of that equation, not an annual survey. Without it, “redeployment” is just a word.

2. Redeployment pathways.

When a process gets automated, where do the people go? If your answer is “we figure it out when it happens,” your first agentic AI rollout is going to be messy in a way that burns trust fast. The pathways — which roles absorb displaced workers, what retraining is required, who owns the transition — need to exist before the automation ships, not after.

3. Workforce modeling that accounts for agents.

Headcount planning that distinguishes between work done by a human, work done by an agent, and work genuinely co-produced. If your workforce planning model still treats this as a headcount question, it is already behind. JPMorgan's flat headcount hides a composition shift of roughly 6% of its roles in a single year. That is the scale you should be modeling for.

4. Human-agent collaboration policy.

This is the newest capability, and it is where most HR functions have the least clarity. It covers policy (who can deploy an agent, with what permissions), governance (audit trails, override protocols, bias testing), and capability (training HR people and line managers to work alongside agents, not around them).

This week's work: Map your current HR function against these four capabilities. Who owns each? Where are the gaps? Where do two partial owners need to become one? Bring the map to your leadership team. That conversation is the first real step. We'll build on it in Issue #02 with the Workforce Redeployment Matrix — a tool for sequencing which processes get automated first and what the redeployment pathway looks like for each one.

· · ·

If your CEO said tomorrow — publicly, the way Dimon did — that AI is already displacing workers in your organization and you have a plan to move them rather than cut them, how confident are you in that plan today? And which of the four capabilities above is the one you'd need to build first to make it true?

Hit reply. I'll share the patterns (anonymized) in Issue #02 and use them to shape the Matrix we build next week.

The Talent Algorithm is an independent weekly brief. Views expressed are those of the author.