Rewiring Talent to Value in the Age of AI (McKinsey)

McKinsey People & Organization article (Korkmaz, Durth, Bérubé; captured 2026-09-27) updating the firm's decade-old Talent to Value framework for an agentic workforce. The headline move: from "Talent to Value" to "Talent and Agents to Value" — the unit of value is no longer a single role but a coordinated system of humans and agents — plus a new fifth step, manage system performance.

Baseline (the original framework)

30–50 critical roles drive ~80% of a company's value. Define a success profile per role, put top performers in them.
Only ~5–10% of those roles report to the CEO (and can deliver ~2.5× the value of an average corporate role); 30–40% sit two levels down, 50–65% three levels below the CEO — i.e. value-critical roles are mostly buried.
McKinsey says the logic "remains highly relevant," but performance gaps between "the best and the rest are widening rapidly."

The five steps, rewired

Define the value agenda — continuously. Value pools shift faster than planning cycles; treat value allocation as a continuous process and redeploy talent and agents toward emerging opportunities. Example: Johnson & Johnson found ~900 gen-AI use cases where 80% of value came from 10–15% of them, and moved from broad experimentation to continuous strategic planning.
Identify critical roles — and agents. Decompose work into automated / augmented / human-led tasks, then decide human vs agent vs hybrid. Roles elevated: domain leaders who own AI-driven outcomes, AI product owners, human–agent workflow architects, prompt engineers who route models, data/knowledge specialists who give agents trusted context. New roles: agentic-workflow mission owners, agent-ops platform leads, agent-governance liaisons — often hired deep in the org and "easily overlooked as critical roles." These cluster in "agent factories" (see Software Factory, Knowledge Work Factory Redesign), so talent pools matter as much as individual roles. Agents themselves must be prioritized like roles (examples: insurance claims processor, pharma patient-trial companion, banking deal-sourcing RM).
Match talent to roles — for AI amplification. Knowledge is commoditized by intelligent systems and experience loses relevance as work changes, so the question becomes how much can this person amplify value with AI? Look for AI "super users" (one person doing what used to take a team) with five markers: builds AI fluency in self and team; reimagines outcomes/workflows through AI; problem framing, creativity and judgment; decision-making and accountability; continuous learning. Example: Meta now defines expected AI-driven impact per critical role and assesses how employees build/use AI tools. Buy / build / borrow — but avoid overreliance on borrowed talent for critical AI capabilities (fragile operating models).
Operationalize — and re-look at the top team. Many exec teams lack the AI fluency to set a value agenda; top teams and boards must be "AI literate, hands-on, and willing to challenge legacy assumptions." Over the past 18 months many Fortune 500 firms reorganized leadership around AI strategy, including executive departures and AI talent taking over existing roles.
Focus on system performance (new). The question shifts from "Who did the work?" to "How well did the system perform?" Two complementary performance models: agents on decision quality, reliability, speed, cost; humans on business impact, ability to define and improve AI-enabled workflows, ethical AI use, cross-team collaboration. Example: a global financial institution evaluates managers on overall workflow performance (throughput, decision reliability, human–agent collaboration, continuous improvement) while agents are scored on speed, accuracy, reliability, compliance. Manage these systems "with the same rigor applied to capital allocation," using data-driven internal talent marketplaces.

Key claims

Failure mode is integration, not talent or tools: organizations "fail to integrate them into coherent systems with clear accountability and feedback loops."
Competitive advantage goes to firms that can "rapidly redesign how work gets done and continuously redeploy talent and AI toward the highest-value opportunities."

Assessment

A consulting-framework piece: directionally strong and well-aligned with the rest of the vault, but light on evidence — the J&J and Meta examples are footnoted public claims, the financial-institution case is anonymous, and there's no data on whether "Talent and Agents to Value" outperforms the original. The value for the vault is the vocabulary (critical agents, agent-ops/governance roles, dual performance models) and the step-5 reframing, which is the most concrete version yet of the "manage the system, not the individual" idea.

Connects to

Talent to Value — the framework page (created from this source).
Advantage Gap — "AI super users" are the Vanguard cohort seen from the HR side; McKinsey's "best vs rest widening" corroborates the gap from a talent lens.
Hourglass Organization — critical roles concentrating in AI-leverage positions; McKinsey's "roles buried three levels down" complements the shape debate.
How AI Is Changing Talent Not Just Tasks (HBR IdeaCast) — same-week HBR take on talent (not just tasks) and where human judgment matters.
AI Engineer (Role) — several of McKinsey's elevated roles (workflow architects, knowledge specialists) are AI-engineering-adjacent.
LLM as Judge / Verification Tax — step 5's agent metrics (decision quality, reliability) need an eval layer to be measurable.

Source

Raw: source — Obsidian Web Clipper (second capture; the first clip carried no body and was purged 2026-09-27).