Growing Without Adding Headcount: Where AI Replaces a Hire and Where It Doesn't

AI rarely replaces a whole role. It takes over tasks: drafting, research, summarising, reformatting and first answers from known facts. Judgement, relationships, accountability and new problems stay with people. So the real question is whether enough of a planned role's tasks can move to AI that the hire can wait, shrink or change.

Can AI replace a hire?

Sometimes it can delay or reshape one, but it seldom removes a whole role. Most jobs mix tasks AI handles well with tasks it handles badly. When enough of the first kind move, the business can grow further before hiring, hire part-time, or hire for a different skill. That is the realistic win.

The headline version, "AI instead of a person", skips the step that decides everything: listing what the person would actually do. A marketing coordinator, an analyst or a sales assistant does dozens of different tasks a month. Some are repetitive writing and data work. Others are judgement calls, follow-ups that depend on trust, and problems nobody has seen before.

So begin with the job description, not the tool. Write down the tasks, the hours each takes, and who checks the output. Only then ask which ones AI could take over with a person reviewing.

Which tasks move to AI, and which stay with people?

Tasks that move are frequent, defined, based on knowledge the business already holds, and quick to check. Tasks that stay depend on judgement, relationships, accountability or physical presence, or carry high risk if wrong. Most roles contain both, which is why AI usually absorbs part of a job rather than all of it.

Research on writing work shows the shape of the shift. In an MIT experiment with professional writing tasks, Shakked Noy and Whitney Zhang found AI mostly substituted for worker effort. It moved work away from rough drafting and towards ideas and editing. In other words, the person stayed and the job changed.

What happened when a company tried to replace a whole team?

Klarna is the most quoted example. In February 2024 it announced its AI assistant was doing the equivalent work of 700 full-time customer service agents. By May 2025 its chief executive was telling Bloomberg the company would invest in human support again, because cost had been weighted too heavily against quality.

According to Klarna's February 2024 announcement, its assistant handled 2.3 million conversations in its first month, two-thirds of its customer service chats. The numbers were real and the volume was impressive.

The correction came within about 15 months. In an interview reported by Customer Experience Dive, chief executive Sebastian Siemiatkowski said cost "seems to have been a too predominant evaluation factor". He added that "there will be always a human if you want". Klarna still uses its AI assistant heavily. The lesson is not that AI failed. It is that the parts of the job involving judgement and reassurance never moved, and customers noticed.

Who gains most from AI: junior or senior staff?

Studies often find newer and less experienced staff gain most, because AI hands them the patterns experienced colleagues already know. That cuts against the idea of replacing junior hires. AI may instead make a junior hire productive sooner, while senior people gain less speed but stay essential for judgement.

According to a study of 5,179 customer support agents by Erik Brynjolfsson, Danielle Li and Lindsey Raymond, an AI assistant raised issues resolved per hour by 14% on average. Novice and low-skilled agents improved by 34%. Experienced and highly skilled agents saw minimal impact.

There is a longer-term reason to keep hiring juniors, too. Today's juniors are tomorrow's seniors. A business that stops hiring entry-level people because AI drafts their work may find, in five years, that nobody has learned the judgement the drafts depend on. The guide to AI SDRs looks at the same question in sales.

How do you decide whether to hire or let AI take the work?

List the planned role's tasks and monthly hours, mark which could move to AI with a person checking, and add up both sides. If the remaining work is less than a meaningful part-time role, delay the hire and measure. If it is more, hire, and give the new person good AI from day one.

Halcyra, a fictional 40-person services firm, planned to hire a marketing coordinator. Its list looked like this:

About 47 of 108 hours could move, counting half the CRM work. The other 61 still needed a person, mostly around events and suppliers. Halcyra hired part-time for those and built two AI workflows for the rest. Then it measured both against a baseline, as the AI ROI framework suggests.

What does the business owe people whose tasks move?

Honesty, a plan for their freed time and training. Tell people early which tasks are moving and why. Point the freed hours at named, measured work rather than leaving them to drift. Train people on the new way of working, and keep them as the approvers of what AI produces.

People notice when work is being redesigned around AI. According to BCG's 2025 AI at Work survey, employees at companies going through comprehensive AI-driven redesign worried more about job security: 46%, against 34% elsewhere. Silence makes that worse. Clear communication about which tasks move, and what people will do instead, makes it better.

Freed time also needs a destination. Hours that disappear into email produce no return, as the article on why time saved is not ROI explains.

Skein starts with the workflow that pays back fastest, usually one the team runs every week, and shows it running before more is committed; the Skein home page explains the approach. That suits a growing B2B business deciding whether its next hire could wait or change shape. A business whose next role is mostly relationships, judgement or physical presence should simply hire, and give that person good AI from the start.

Frequently asked questions

Can AI replace an employee?

Rarely a whole one. Most roles mix tasks AI handles well, such as drafting and research, with tasks it handles badly, such as judgement and relationships. AI usually absorbs part of a job, which can delay a hire, make it part-time or change the skills needed, rather than remove it.

Which jobs is AI most likely to replace?

AI is most likely to take over tasks rather than jobs: routine drafting, summarising, reformatting, data entry and first answers from known facts. Roles made up mostly of those tasks change the most. Roles built on relationships, accountability and new problems change least, though the people in them still use AI.

Should a small business use AI instead of hiring?

Only after listing the planned role's tasks and hours. If most of the work could move to AI with a person checking, delay the hire and measure. If the remaining work still needs a person, hire, and give them good AI from the start so they are productive sooner.

Does AI help junior or senior employees more?

Studies often find less experienced staff gain most. A study of over 5,000 support agents found novices improved by 34% with an AI assistant, while experienced agents saw minimal impact. That suggests AI can make junior hires productive faster rather than make them unnecessary.

How do you tell staff that AI is taking over some of their tasks?

Tell them early, explain which tasks are moving and why, and say what the freed time will go towards. Train people on the new workflow and keep them as the approvers of what AI produces. Silence breeds worry about job security; a clear plan reduces it.

Related reading