6 new CTO roles in the AI era

Just 3 years ago, the CTO mostly dealt with infrastructure, the dev team, and vendors. Today? The CEO asks about AI strategy, the CFO about ROI from models, HR about the impact on headcount. According to McKinsey, companies that win with AI have technical leaders who rebuild processes from the ground up. They do it 3x more often than the rest. The CTO role has changed.

1. AI Translator

The business speaks one language, technology another. The CTO has to translate both ways. "We want better customer service" → "We need RAG with a product knowledge base." And vice versa.

This is not optional. If you don't translate, the vendor will, and they have their own interests. A model vendor is writing about work and power too. What survives a check of it is in OpenAI's essay on a state that stops depending on its citizens' work.

2. AI Portfolio Manager

Not one AI project. A portfolio of projects with different horizons and risks. The CTO has to manage it like an investment fund:

AI PORTFOLIO

Quick wins (60%): Low risk, fast value, build momentum
Strategic bets (30%): Medium risk, bigger impact, 6-12 months
Moonshots (10%): High risk, potentially transformative

3. AI Risk Officer

Who in the company understands the risks tied to AI? Hallucinations, bias, privacy, security, compliance? In most companies, nobody. By default it lands on the CTO.

You don't have to be an expert in everything. But you do have to know which questions to ask and who to hire.

Which questions those are, and what structure catches model mistakes before they cost money, I laid out in a separate piece on AI governance. It opens with the EY survey: 99% of companies booked losses from AI risks.

4. AI Capability Builder

Your team has no AI competencies? That's your problem. You have three options:

• Build: Train your existing team, hire specialists
• Buy: Outsource to firms specializing in AI
• Partner: Work with vendors who have the competencies

Most companies need a mix of all three.

The "build" path, meaning three levels of readiness and the training program behind them, I broke down in a post on preparing your team for AI.

"Your strength as a CTO is building a team that knows AI better than you do."

WORK WITH ME

This is what I do hands-on: advising on AI strategy and building agents that survive the demo.

AI consulting AI implementation

5. AI Ethics Guardian

Who decides whether to use AI to evaluate employees? Can AI reject candidates? Who is accountable when the model discriminates?

These questions land on the CTO's desk. And you can't answer "that's not my department."

The decision stays with you, but the work behind it can be delegated to an AI Ethics Lead or a five-person committee.

6. AI Change Agent

Technology is 30% of success with AI. The rest is organizational change. People fear for their jobs, processes have to change, the culture has to evolve. That change starts in the departments that do not write code, and its first step is often a single day of training. How to run a first AI session for people outside IT so that somebody uses it on Monday, I wrote up from a newsroom.

A CTO who rolls out AI without managing change is rolling out a failure.

What to stop doing

To take on new roles, you have to delegate something:

DELEGATE OR AUTOMATE

✗ Micromanaging daily sprints
✗ Being the only person who can approve a PR
✗ Negotiating with every vendor personally
✗ Writing code (yes, I know you love it)

How to prepare

1. Spend time with AI. Not theoretically, practically. Build something.
2. Talk to the CEO in the language of business, not technology
3. Build a network: other CTOs, AI experts, advisors
4. Read the regulations: EU AI Act, GDPR, industry-specific
5. Practice making decisions under uncertainty

What to do this week

Pick the one of these six roles where your gap is biggest. This week, do one concrete thing in it: build a small prototype, call another CTO, or read a section of the EU AI Act. The 2020 CTO role is not coming back, and nobody hands you a manual for the new one.