You're not behind. You're just starting.

If you haven't really used AI tools yet, you probably feel like you've missed the boat. Everyone on LinkedIn has apparently "transformed their workflow with AI" and you're still doing things the old-fashioned way, feeling vaguely guilty about it.

Relax. Most of those LinkedIn posts are performative nonsense written by people who asked ChatGPT to write a LinkedIn post about how they use ChatGPT. The irony is thick enough to spread on toast.

The truth is, you're actually in a decent position. The tools have matured significantly since the initial hype wave. The genuinely useful stuff has separated from the gimmicks. You get to skip the "paying $40/month for a tool that barely worked" phase that early adopters suffered through.

Let's get you started properly.

Try these three things right now (seriously, right now)

Before we get into theory and comparisons, I want you to actually experience what these tools can do. Open a new tab and try each of these. It'll take about ten minutes total. Go to chatgpt.com and ask it something you've been curious about.

Not "write me an essay." Ask it something you'd normally Google but never get a satisfying answer for. Something like: "Explain how mortgages actually work in the UK. I understand the basic concept but I don't understand what happens between applying and getting the keys. Walk me through it step by step." Or: "I'm thinking of learning to play guitar at age 35. Is it realistic and where should I start?"

Notice how different this feels from a Google search. You get a direct, conversational answer instead of ten blue links to websites that may or may not answer your actual question. Go to claude.ai and paste a paragraph of your own writing.

Take something you've written recently, an email, a report, anything. Paste it in and say: "Help me make this clearer and more concise without changing the meaning or my voice." Watch what comes back. It should feel like a really good editor suggested improvements, not like someone rewrote your text in robot-speak. Go to perplexity.ai and research something.

perplexity is an AI-powered research tool. Ask it something factual that you'd normally have to read multiple articles to understand. Something like: "What are the current best options for a savings account in the UK and how do the interest rates compare?" It'll give you a sourced, up-to-date answer with citations you can verify.

If you tried all three, you now have a better understanding of AI tools than most of the people posting about them on social media. Let's build on that.

What AI tools actually are (in plain English)

AI tools like chatgpt and claude are essentially very sophisticated text prediction systems. They've been trained on enormous amounts of text from the internet, books, and other sources. When you type a message, they predict what the most helpful response would be, word by word.

This is both their superpower and their biggest limitation. They're incredibly good at anything that involves language: writing, summarising, explaining, translating, brainstorming, analysing text. They're unreliable at anything that requires being factually correct about specific details, because they're predicting plausible text, not looking things up in a database.

Think of them less like a search engine and more like a very well-read colleague. Smart, articulate, helpful, but occasionally confidently wrong about specific facts. You'd trust them to help you draft a presentation, but you'd double-check any statistics they cite.

The myths, debunked

Myth: AI is going to take your job tomorrow. Reality: AI is changing how some work gets done. Some routine tasks are being automated. Some jobs will evolve. But the "everyone will be unemployed in two years" crowd has been wrong about every technology prediction they've ever made. If your job involves judgement, creativity, human relationships, or physical presence, you're fine. If your job is entirely "take this input and produce this standard output," that's the part AI is good at.

Myth: AI can think. Reality: It really, genuinely cannot. It processes patterns in text. It has no understanding, no awareness, no opinions. When Claude says "I think this approach would work well," it's generating text that pattern-matches to a helpful response, not expressing a thought. This matters because it means you should treat AI output as a starting point, never as a final answer.

Myth: AI makes things up constantly and can't be trusted. Reality: The "making things up" problem (called hallucination) is real but overstated. Modern models like GPT-4o and Claude Opus are dramatically more reliable than the models from 2023. They still hallucinate, especially about specific facts, dates, and names, but for most practical tasks (writing, brainstorming, explaining concepts), the output is solid. Just don't trust specific statistics without checking them.

Myth: You need to learn "prompt engineering" to use AI effectively. Reality: Just talk to it like a person. Seriously. The best "prompt" is usually just a clear description of what you want, written the way you'd explain it to a smart colleague. "Write me a professional email declining a meeting invitation. I want to be polite but firm. The meeting is about Q3 planning and my reason is that I have a conflicting deadline." That's all the "prompt engineering" most people need.

The tools worth knowing about

There are hundreds of AI tools. You need to know about five. Maybe six.

chatgpt (by OpenAI) is the one everyone's heard of. It's a general-purpose AI assistant that can write, code, analyse data, browse the web, and generate images. The free tier is solid. The Plus plan at $20/month unlocks the best model and higher usage limits. This is the default recommendation for most people.

claude (by Anthropic) is ChatGPT's main competitor. It's particularly good at writing that sounds natural, handling long documents, and coding. Many people (myself included) find Claude's responses feel more thoughtful and less formulaic than ChatGPT's. The free tier is decent; Pro is £16/month. Worth trying even if you end up preferring ChatGPT.

perplexity is what Google should have become. It's an AI-powered research tool that answers questions with sources. When you need factual, up-to-date information, Perplexity is more reliable than ChatGPT or Claude because it actively searches the web and cites its sources. Free tier is excellent.

canva has evolved from a simple design tool into an AI-powered creative suite. If you ever need to make a presentation, social media graphic, or simple design, Canva's AI features make it embarrassingly easy. Free tier covers most needs.

grammarly is a writing assistant that checks your grammar, spelling, tone, and clarity. The AI features can rewrite sentences and suggest improvements. The free tier catches basic errors; Premium adds the smart suggestions. Worth having whether or not you use any other AI tool.

How to get the most out of AI tools (practical tips)

Be specific about what you want. "Help me write an email" is vague. "Help me write a polite but firm email to my landlord about a broken boiler that's been unfixed for two weeks" gives the AI enough context to be genuinely useful.

Tell it about your context. AI doesn't know who you are, what you do, or what you've already tried. The more relevant context you provide, the better the output. "I'm a secondary school teacher writing to parents about a field trip" gets much better results than "write a letter about a field trip."

Iterate, don't accept the first output. The first response is rarely perfect. Say "good, but make it shorter" or "I like the structure but the tone is too formal" or "can you include a specific example?" Treat it like a conversation, not a vending machine.

Use it for what it's good at. First drafts, brainstorming, explaining concepts, summarising long documents, generating options, checking your work. Don't use it as a source of truth for specific facts, medical advice, legal advice, or anything where being wrong has serious consequences.

Don't overthink it. The best way to learn what AI tools can do is to use them regularly for small tasks. Next time you need to write an awkward email, try it. Next time you need to understand a complex topic, ask it. Next time you need ideas for something, brainstorm with it. The understanding comes from use, not from reading articles about it. Including this one.

Your first week plan

Monday: Sign up for ChatGPT (free) and Claude (free). Ask both the same question and see which response you prefer. Tuesday: Use your preferred tool to help with a real work task. Draft an email, summarise a document, brainstorm ideas for a project. Wednesday: Try Perplexity for a research question you actually care about. Compare the answer quality to a regular Google search. Thursday: Use an AI tool to explain something you've always been curious about but never had time to properly research. Friday: Reflect on which moments felt useful and which felt like a gimmick. That's your guide for what to keep using.

That's it. No courses, no certifications, no "prompt engineering masterclass." Just start using the tools for real tasks and you'll figure out where they fit in your life naturally.