AI is splitting the growth model: why executive teams need a new learning and talent strategy
The idea that learning ends after a degree is outdated. That's the message from Hemant Taneja (General Catalyst) and Bob Sternfels (McKinsey) - and the numbers back it up.
On stage at CES 2026 during a live taping of the All-In podcast, Taneja put it bluntly: "This idea that we spend 22 years learning and then 40 years working is broken." In an era where AI agents can be trained faster than employees, the bar for staying relevant keeps rising. As host Jason Calacanis said: "You're going to have to show chutzpah, drive, passion."
What McKinsey's operating data signals for your org
McKinsey has used AI to increase client-facing roles by 25% and reduce non-client-facing roles by a similar amount - while lifting total output by 10%. The firm expects to have roughly as many AI agents as human employees by year-end; today, humans outnumber agents 40,000 to 25,000.
Sternfels summed it up: growth used to mean more people everywhere. Now it splits. You grow on the client side, shrink elsewhere, and still expand overall.
AI job anxiety is real - especially for early careers
Young workers have felt the first shock. Research cited from Stanford's Digital Economy Lab points to a 13% relative decline in employment among 22-25-year-olds in high-risk occupations since 2022. Gallup has reported that more than a third of workplaces already use AI.
Still, don't torch the entry-level pipeline. Even with heavier AI adoption across Amazon, AWS CEO Matt Garman called replacing junior roles "one of the dumbest things I've ever heard." No pipeline, no future leaders.
Executive implications: where to reallocate attention and budget
- Adopt skills-based talent models: Map critical skills by role, not degrees. Fund recurring re-skilling cycles and make them part of annual planning, not a side project.
- Plan "AI agent headcount" alongside human headcount: Set targets for agent-to-employee ratios, define ownership, and track output per combined workforce (people + agents).
- Rebalance the org: Shift capacity from back office to revenue and client impact. Cross-train operations talent into client-facing, data, and automation roles.
- Build a learning operating system: Short, continuous sprints tied to live projects. Prioritize on-the-job use cases over passive courses.
- Update incentives: Reward shipped automations, skill acquisition, and certifications. Tie compensation to measurable productivity gains, not seat time.
- Protect the junior ladder: Use AI to augment apprenticeships and rotations, not to delete them. Managers should own the pipeline health metric.
- Governance and risk: Create clear guardrails for data, model use, and approvals. Keep humans accountable for decisions AI informs.
- Communicate the change: Publish role transitions, re-skilling paths, and timelines. Transparency lowers anxiety and speeds adoption.
- Instrument the metrics: Track time-to-productivity, automation hours saved, revenue per employee+agent, client NPS, and learning completion/impact.
A 90-day plan to operationalize continuous learning
- Days 0-30: Identify the top 5 workflows per function for AI augmentation. Define skill gaps. Select a small set of tools and secure data access.
- Days 31-60: Launch hands-on sprints tied to those workflows. Pair every team with a "use case owner." Start publishing weekly automation wins.
- Days 61-90: Scale what works. Fold AI agent metrics into performance reviews. Lock a quarterly re-skilling cadence and budget.
Strategic takeaways for leadership
- Learning is now a permanent function, not a perk. Treat it like infrastructure.
- AI shifts the center of gravity to client impact. Staff and measure accordingly.
- Entry-level roles are your bench. Keep them, augment them, grow them.
Get your team moving
If you need a curated path for role-based AI upskilling, explore practical programs built for on-the-job use:
The pace isn't slowing. As Taneja put it, the old "learn, then work" model is done. Build a company that learns while it works - and works better because it learns.
Your membership also unlocks:
