The rise of the AI-augmented employee

The Wrong Conversation

The public debate about AI and employment has been stuck in a binary: either AI replaces jobs, or it doesn't. Both sides produce statistics, studies, and projections. Neither is particularly useful for a business leader trying to make decisions today.

The more accurate — and more actionable — framing comes from MIT economist David Autor, who has been studying labour and technology for 30 years. His thesis: automation doesn't eliminate jobs, it eliminates tasks within jobs. The job then reconstitutes around the higher-complexity tasks that remain.

This happened with accounting software, CRM platforms, and document management. It's happening again with AI — faster and more broadly.

What "Task Replacement" Looks Like

Consider what a mid-level analyst at a financial advisory firm actually does in a week:

• —Pulls data from Bloomberg and internal systems: ~6 hours
• —Cleans and formats data into presentation templates: ~4 hours
• —Writes first-draft commentary sections: ~3 hours
• —Attends client calls and takes notes: ~4 hours
• —Summarises notes and drafts follow-up emails: ~2 hours
• —Does actual analysis — pattern recognition, hypothesis formation, recommendation building: ~5 hours

The first three items — roughly 13 hours out of 24 — are now automatable with off-the-shelf AI tooling. The last three involve judgment, relationships, and reasoning that AI meaningfully assists but doesn't replace.

The analyst who learns to delegate the first three to AI and spend 18 hours on the last three becomes dramatically more productive. The analyst who doesn't gets managed out of a role that now requires fewer people to do.

The Management Implication

This is where most leaders are underinvesting: not in AI tools, but in deliberate task-level redesign.

Most organisations are running informal experiments. Someone in finance is using ChatGPT to draft memos. Someone in operations is using Claude to summarise meeting transcripts. These individual productivity gains are real but they don't compound — because they're not embedded in the workflow.

The companies pulling ahead have done something more systematic: they've gone through key job functions and explicitly mapped out:

• —Which tasks are automatable today with high confidence
• —Which tasks AI can do a draft of, requiring human review and refinement
• —Which tasks require human judgment and should get more time as a result

Then they've built the tooling, prompts, and processes to make the first two categories actually happen — and they've adjusted performance expectations for the third.

This isn't a technology project. It's a management and design project that happens to involve technology.

What We're Seeing

Across the clients we work with — primarily companies between 20 and 500 employees — a few patterns are emerging consistently:

Customer-facing roles are being augmented fastest. Support, sales, and account management teams are getting AI tools for drafting, summarising, and researching — and the best performers are using them to take on more accounts or handle more complex issues, not to work less.

Back-office functions are being restructured. Finance, HR, and legal teams are finding that AI handles the routine extraction, classification, and generation tasks well enough to change headcount ratios. A team that needed 6 people to process a certain volume of work now needs 3, with the remaining 3 handling more complex work or being redeployed.

Middle management is under the most pressure. The coordination, summarisation, and information-relay functions that a lot of managerial work involves are exactly what AI does well. Companies are finding they need fewer layers of reporting relationships when information flows more freely.

How to Start This Conversation

If you're a CEO or business leader, the most useful thing you can do right now isn't to pick an AI vendor. It's to sit down with your team leads and have a structured conversation about tasks — not jobs.

Ask: "If we could offload the most repetitive 20% of your team's work to an AI assistant that was good but not perfect, what would change?"

The answers will tell you where to invest. They'll also tell you something about which team leads are thinking about this clearly and which ones are still in denial.

The job market for people who can work effectively alongside AI is already diverging from the market for those who can't. The same divergence is coming for businesses.

The companies that get ahead of this won't necessarily have the best AI tools. They'll have the best habits of integrating AI into actual work — and those habits are built intentionally, not by accident.

What This Means for Hiring

One practical implication worth calling out: the unit of hiring is shifting from role to capacity.

A year ago, you might hire a content writer to produce 4 pieces per week. Now, a strong content writer using AI tools can produce 8-12 pieces per week at the same or better quality. You don't need to hire two writers — you need to hire one writer who knows how to use the tools, and pay them more.

The same logic applies across functions. Expect your high performers to become much higher performers as they adopt AI tools. Expect the gap between your best and average performers to widen. And expect the total headcount required for a given level of output to decrease — not dramatically, and not all at once, but measurably over the next 2-3 years.

Plan for it now. The companies that don't will be forced to react.