The second path into an AI role — a checklist · Resource

This is the version you keep on hand. Open it when you want to take the second path from the article: becoming the AI-savviest person at the company you already work for, instead of starting an agency. The steps run in order: first name the terms and check the assumptions, then close the gap in your own role, then hold course and watch the traps.

The governing rule in one sentence: you're not changing careers, you're changing which version of your career you are.

0. Before you start — what you need to know

Three terms run through the whole path. Fix them in your head before you go further.

AI automation agency — a firm that implements AI-based solutions for other businesses, most often automating repetitive processes.
Agent — a program that carries out a multi-step task on its own (reads a ticket, finds an answer, drafts a reply) instead of just answering one question.
Implementation — taking a tool from idea to daily use inside a company.

Also take on two assumptions about the numbers you'll be citing:

[ ] Remember that the data comes from an IBM survey of 2,000 CEOs of large public companies, with median annual revenue of around $5.8 billion. These are big, established firms, not a global average.
[ ] Assume the global rate is clearly lower — the article puts the estimate below 30 percent.

1. Understand why there's more than one chair

First see what actually changed in company structure, without which the next steps don't make sense.

[ ] Note that 76 percent of the CEOs in the survey either already have a chief AI officer (CAIO) or are hiring one this year, up from 26 percent two years earlier.
[ ] Don't aim only at the chief AI officer's seat: the same survey says every other member of the leadership team, from marketing to finance to operations, is expected to become AI-fluent too.
[ ] Treat the new role like the chief information security officer (CISO) after the internet arrived: a new problem showed up, so a new role got created. This time it took about 24 months, not a decade.
[ ] Keep the takeaway in mind: one chair is the most visible, but there are far more open seats than that.

2. See the gap you need to close

The article's second number shows where the bottleneck sits, and what you'll actually be doing.

[ ] Take on two numbers from the surveyed companies: only 25 percent of employees actually use AI tools in daily work, while CEOs say 86 percent of their people either have the right skills or would pick them up after brief training. That's a 61-point gap.
[ ] Weigh the author's caveat: this is survey data, and employees under-report what they actually use, so the exact size of the gap is uncertain, but the gap itself is real.
[ ] Name your job plainly: connect people to the processes that genuinely need this — build the bridge between "we know how to use it" and "we're saving forty hours a week because of it."
[ ] Expect the bridge not to build itself: change management can be exhausting, because the short-term cost (training people, taking processes apart and putting them back together) comes before the long-term payoff.

3. Pick a path — and don't wait for the title

The article lays out two paths to the same place. Decide which one you're taking, but start acting regardless of the choice.

[ ] Path A — from outside. Start as a consultant or launch an AI agency, take on a few clients, solve their problems; some of them will eventually hire you full time.
[ ] Path B — internal promotion. You already have a job: quietly become the most AI-fluent person in the building, bring ready-made prompts to meetings, build small automations out of curiosity, and become the obvious candidate when the new role opens.
[ ] Don't assume path A is the only one: a separate IBM survey of 600 chief AI officers found that 57 percent were promoted from within the company, doing the job before it even had a title.
[ ] Start closing the gap now, in your current role. Neither path requires waiting to be hired for it.

The concrete first move the article spells out:

Pick one process in your area that nobody on the team has touched with AI tools yet.
Build an AI-assisted version of it.
Document how much time it saves.
Show it to your manager.

4. Build around what you already like

According to the article, this is the most important piece of the puzzle: you can't stick with something you don't enjoy.

[ ] Build the AI-assisted version of what you already do and like: if you like marketing, automate content, copy and pages instead of forcing yourself to build finance agents.
[ ] Aim to become the AI-fluent specialist in your field, the one your manager eventually promotes.
[ ] Don't try to be someone else: according to the article, that's a common source of impostor syndrome among people learning AI.
[ ] Lean on what the CEOs themselves say: 85 percent say every functional manager needs to become a technology expert, and 77 percent think the "people" and "technology" roles are merging — the winner is whoever combines both.
[ ] Keep the internet analogy in mind: today there are "AI consultants," but soon they'll just be consultants, and whoever isn't AI-fluent won't keep up with the rest.

5. If you work in a regulated industry

The article carves out healthcare, finance and the public sector separately, where you can't just plug AI into company data.

[ ] Don't assume this doesn't apply to you: industry knowledge combined with AI fluency under real constraints is one of the rarest profiles on the market today.
[ ] If the answer is "no" for now, build your own projects after hours, on substitute data.
[ ] Show those projects to your team, so that when the company gets the green light, your name is the first one that comes up.

Watch out for

Two honest notes the article uses to temper enthusiasm for the numbers.

The numbers come from large, established companies. Most of us don't work at companies like that, and the global rate is probably clearly lower. Don't map them directly onto your own situation.
CEOs can be far off in their forecasts. In 2024, half of them thought AI would be the main growth driver by 2026; today only 10 percent say that. That's a roughly 40-point miss in a year, so treat predictions with caution.

What the article treats as not up for debate: CEOs are hiring, company structure is changing, and the functional managers who get promoted are the ones who know AI best.

The conclusion is short: you don't need to change careers, you need to change which version of your career you are. Thinking can be outsourced; understanding can't.