Leadership in the age of AI

— it's in your hands

1

Protect the time

notes
ideas
practice
learn
reflect
grow
2

Invest in the tools

explore
try
ask
iterate
build
test
3

Apply it to your actual job

refine
check
verify
adapt
improve
repeat
4

Hone asking and checking

share
teach
lead
inspire
change
thrive
5

Build your team's capability

Ready to start?

You can't delegate this one.

For leaders in the public sector, business, and startups who want to learn AI themselves — not delegate it to someone else.

You've built a career on commissioning, briefing, steering. Handing technical stuff to people who know it better. Trusting managers to work out the how.

That won't work with AI.

When spreadsheets arrived, you could hire someone who knew Excel. AI is different. It's not a tool you hand to a specialist. It's a new language for leading — and you can only learn it by speaking it yourself.

The good news: you already know how to ask good questions. You brief executives and ministers, steer boards, challenge consultants. That's the core skill. AI just needs you to practise it in a new way — every day, on real problems, until it becomes instinct.

No course teaches this. No consultant can do it for you. The leaders pulling ahead aren't more technical. They're more disciplined about practising.

Ten minutes a day. That's all it takes to start.

What's "vibe coding"? It's the new way of building things with AI — you describe what you want in plain English, AI writes the code, you test and refine. Collins Dictionary's word of the year 2025. No programming skills required. Just clear thinking and persistence.

Start the 5 steps
Step 1

Find 10 minutes a day — and defend them

This is the hardest step. Not because AI is complicated, but because your diary is block-booked.

You already know how to learn. You've picked up new skills throughout your career — often under pressure, often without formal training. AI is no different. What's different is that no one is going to give you time to do it.

So take it.

Block 10 minutes. Every day. Same time if you can. Treat it like a meeting you can't cancel — because for the next few months, it's the most important one you have.

Why 10 minutes? Because it's short enough to actually happen. Long enough to try something real. And it compounds. Ten minutes a day for a month is five hours of practice. In six months, you'll have built a skill that most leaders are still delegating.

Try this now

Open your calendar. Block 10 minutes tomorrow morning. Call it "AI practice" or "thinking time" — whatever gets it past the meeting requests.

Using AI to protect your time

Already have access to AI at work? Use it to book the time.

Microsoft 365 Copilot

Google Workspace / Gemini

Step 2

Get set up properly — your own subscription, your own space

Follow your work's IT policies. If you're using a personal subscription, don't give it problems that are owned by your employer. Keep work data on work-approved tools.

If you're using the free version of ChatGPT, or waiting for IT to roll something out, you're not serious yet.

This isn't about money — the best tools cost less than a newspaper subscription. It's about commitment. When you pay for your own account, you own the space. No one's watching your prompts. No one's judging your questions. You can experiment, fail, and learn.

Get a paid subscription to Claude, ChatGPT, or Gemini. All three are good. Claude is particularly strong for writing and analysis. ChatGPT is the most widely used. Pick one and stick with it for a month before comparing.

Don't wait for your organisation's enterprise solution. By the time it arrives, you'll have missed months of learning — and you'll be behind the people who just got on with it.

Set up a dedicated space for learning

Consider setting up a new Gmail account specifically for your AI learning. This keeps your experiments separate from work, lets you fully integrate with AI tools without corporate restrictions, and gives you a clean space to build your skills.

Getting started with GitHub

If you want to try vibe coding — building tools by describing what you want — you'll need a GitHub account. GitHub is where code lives. You don't need to understand code to use it.

Try this now

Sign up for a paid AI subscription. £18/month. Cancel any time. You'll know within a week whether it's worth it.

Step 3

Use it for something real — today

Only use AI tools that your organisation has approved. If you're in the public sector, check your IT policies before using any AI on work problems. Many councils and departments now have approved tools — use those.

The fastest way to learn is to solve a problem you actually have. Not a test. Not a demo. Something from your to-do list that you'd normally spend an hour on.

Start with something you do every week. A briefing note. A submission. A response to a parliamentary question. A summary of a consultation.

Paste in the source material and ask for a first draft. You'll learn more in one real task than in ten tutorials — because you'll see where it helps, where it hallucinates, and where your judgment still matters.

Public sector leaders — whether in central government, local authorities, the NHS, or arm's-length bodies — are using AI to summarise long documents for briefings to executives, ministers and elected members, draft options appraisals, and prepare speaking notes for committees. The UK government's AI Playbook has examples of what's working.

The question isn't whether AI can do your job. It's whether you can direct it well enough to make your job easier — while keeping the accountability where it belongs.

Try this now

Take a document you need to summarise this week. Paste it in. Ask for a one-page brief. See what you get.

Step 4

Get better at asking — and better at checking

You already know how to ask good questions. You brief executives, ministers and elected members, steer boards, challenge consultants. That's the core skill. AI just needs you to practise it in a new way.

The better your prompt, the better the output. Be specific. Give context. Say what you want and what you don't want. If the first answer isn't right, don't start again — refine the question.

Think of it as a dialogue, not a search

The biggest mistake new users make is treating AI like Google — one question, one answer, done. That's not how it works. The real power comes from conversation.

When you get a response, interrogate it. Ask "Why did you suggest that approach?" or "What are the risks with this?" or "Give me three alternatives." Push back. Ask for sources. Request it in a different format. Say "That's too long — give me the three most important points."

You wouldn't accept the first draft from a junior colleague without questions. Don't accept it from AI either. The dialogue is where the value emerges — and where you learn what works.

Checking is the essential skill

AI will confidently present information that's wrong. It will invent sources. It will miss nuance. Your job is to check — not to trust. This is called hallucination, and it happens more often than you'd expect.

Build checking into your workflow. When AI gives you statistics, verify them. When it cites a report, find the original. When it makes a recommendation, ask yourself what it might have missed. This isn't busywork — it's the difference between using AI well and being embarrassed by it.

This is the skill that separates useful AI from dangerous AI. And it's a leadership skill, not a technical one. You've spent your career knowing when to trust an answer and when to dig deeper. Apply that here.

This is creative work

Here's what surprises people: using AI well is genuinely creative.

You're not outsourcing your thinking — you're freeing it. The boring bits get faster. The interesting bits get more interesting. When you're not spending an hour formatting a document or drafting boilerplate, you have time for the work that actually needs a human brain: the strategy, the judgment, the "what if we tried this?"

The best AI users describe a loop: describe what you want, see what comes back, think about what's wrong with it, refine, repeat. That's not mechanical. That's creative problem-solving. It's the same skill you use when you're shaping a strategy or challenging a proposal — just applied in a new way.

Leaders who get this right don't feel like they're working with a machine. They feel like they're thinking more clearly than before.

Try this now

Take something you asked AI for recently. Don't just read the answer — have a conversation with it. Ask follow-up questions. Challenge its assumptions. Check one fact. Get in the habit of dialogue, not just query.

Step 5

Bring your team with you — not just yourself

Learning AI isn't a personal project. It's a leadership responsibility. Once you've got the basics, start sharing — not by mandating training, but by modelling the behaviour.

Your team is watching what you do, not what you say. If you're still delegating AI to "digital" or waiting for the enterprise solution, they'll do the same.

The civil service "One Big Thing" programme is asking all staff to complete AI learning by February 2026. But workshops create awareness, not capability. Real skill comes from practice in context, timely feedback, and repetition.

Create permission to experiment. Share prompts that worked. Ask what your team has tried. Make it safe to fail. The organisations pulling ahead aren't the ones with the best AI strategy documents — they're the ones where leaders are visibly learning.

This also means being responsive. Governance matters, but so does speed. If your team needs six months of approvals to try a new tool, you've already lost ground to departments that moved faster.

Try this now

In your next team meeting, share one thing you've learned. A prompt. A use case. A mistake. Make it normal.

Keep learning

Podcasts

Hard Fork (New York Times) — Kevin Roose and Casey Newton on tech, AI, and the forces shaping our world. Accessible, funny, current.

Exponential View (Azeem Azhar) — Deep dives on AI, leadership, and exponential change. Essential for understanding where this is heading.

Boss Class (The Economist) — Management lessons and leadership insights. Increasingly focused on how AI is reshaping how leaders work and make decisions.

Further reading

Follow along

I'm sharing what I learn as I go — prompts that work, mistakes I make, tools I try.

@vibe2lead on Bluesky

Glossary

Vibe coding

Describing what you want software to do in plain English, letting AI build it, then testing and refining. You don't need to understand the code — you just need to know what you're trying to achieve. Named Collins Dictionary's word of the year for 2025.

Back to where you were
Prompt

The question or instruction you give to an AI. The better the prompt, the better the answer. Good prompts include context, constraints, and examples of what you're looking for.

Back
Hallucination

When AI confidently presents information that's wrong or made up. Common with names, dates, sources, and statistics. It's not lying — it's generating plausible-sounding text without checking facts. Always verify anything important.

Back
Large language model (LLM)

The technology behind tools like ChatGPT, Claude, and Gemini. Trained on vast amounts of text to predict useful responses. Think of it as a very sophisticated autocomplete that can understand and generate human language.

Back
Agentic AI

AI that can take actions on its own — booking meetings, running code, browsing the web — rather than just answering questions. Still emerging, but coming fast. The next wave after chatbots.

Back
Context window

How much information an AI can "hold in mind" at once. Bigger windows mean you can paste longer documents or have longer conversations before it forgets the beginning. Measured in "tokens" (roughly words).

Back

About

I'm Vasant. I lead teams and set strategy. I don't have any more time than you — most of this was built on trains, with a chatbot and ten minutes of stubbornness. If I can do it, you can.