What To Pay For And What To Skip With AI
What the free plan is for, what the paid one adds, and a prompt that decides whether you should upgrade this month or keep your £20.
The most profitable person in AI right now is a nervous beginner. By the time they’ve written a decent prompt, someone has sold them two subscriptions, a “1,000 prompts” PDF and a course with a countdown timer.
Working for yourself makes it worse. Every one of those is a business expense you can talk yourself into.
So: what the free plan is for, what the paid one adds, and how to decide.
What the free plan is for
The bit the ads never mention: Claude’s free plan is fine for trying Claude. Real conversations, writing, planning, thinking a problem through with it. Most beginners can learn the whole skill of asking well without paying a penny.
The catch is volume. Free comes with a usage limit, and when you hit it Claude asks you to come back in a few hours. Nothing is broken. What to actually do with it: The Free AI Setup That’s Genuinely Enough.
What the paid plan is for
Pro, around £20 a month, is where the setup stuff lives:
- Projects, a place per client where the context stays put.
- Connectors, so Claude can read your Gmail, Drive, Calendar or Notion instead of you pasting.
- Skills, so it follows your way of doing a job without being told every time.
- Scheduled tasks, Claude doing something on a timetable without you opening it. A brief waiting at 7am rather than one you have to ask for.
- Bigger usage limits, for the days you’d otherwise hit the cap mid-job.
What sits on which plan shuffles over time, so check Claude’s plans page on the day.
Rule 1: stay free until it hurts
Free is your actual plan, not a trial. Use it daily and let it get in your way before you reach for a card.
Hitting the limit is information, not punishment. Most beginners never touch it, which tells you free fits. Hitting it repeatedly during paid work tells you the other thing. That, or wanting the setup features above for real work, are the 2 upgrade signals worth trusting. The ads, the FOMO, the “serious people pay” energy: noise.
Rule 2: one subscription, never a stack
When the day comes, upgrade one tool. The one you use.
Claude’s paid plan is priced in US dollars at around twenty a month, landing near £20 once your card and the VAT have had their say. Check the site on the day, that number moves.
Two of those is roughly £480 a year for one job done twice. The stack comes from headlines: every month some model is newly “the best” and the twitchy move is to subscribe to the winner while keeping the old one just in case. The differences at the top are small and keep swapping, and your settled setup, with your memory and your projects in it, beats a marginally cleverer stranger. If you genuinely switch, cancel first, then subscribe. Swap, never stack.
Rule 3: never pay for what’s free inside the tool
Three I wouldn’t spend a pound on.
Prompt packs. Ask the tool instead. “Write me a great prompt for quoting a job I’ve never quoted before” beats the pack, tailored to you, for nothing. Buying prompts is buying bottled water while stood in a stream.
Funnel courses. The £297 to £997 kind, with the timer and the income screenshots. Inside is usually the content of free guides, read aloud over slides. Not all paid learning is a con, but good teaching doesn’t need a fake clock. The clock is the tell.
Wrapper apps. Anything charging monthly whose main feature is being a nicer window onto a model you could use directly. Close the tab.
All three sell the same feeling: everyone else knows something you don’t. They don’t.
The decision prompt
Don’t decide this mid-annoyance. Make Claude walk you through it.
Nick this one:
You're helping me decide whether to upgrade to a paid AI plan this month, or keep the money. I'm self-employed, so this comes out of the same pot as everything else.
How I've been using it: [e.g. "most weekdays, drafting client emails and summarising receipts"]
How long: [weeks or months]
What happened when I hit the free limit: [e.g. "twice last week, mid-invoice-run on deadline day, finished by hand at 11pm"]
AI things I already pay for: [subscriptions, apps, courses, or "nothing"]
Ask these one at a time, waiting for each answer:
1. How many times have I hit the limit in the past fortnight?
2. Was I doing paid work when it happened, or having a play?
3. What did it cost me: a task abandoned, an evening rearranged, work done worse by hand?
4. Am I using it most days or most weeks?
5. Do I need the setup side (Projects, connectors, Skills, scheduled tasks), or do I just need more room?
Then give me:
1. A straight recommendation: upgrade now, wait two weeks and count again, or stay free.
2. The reasoning, in three lines, based only on what I told you.
3. Anything I'm already paying for that I should cancel first.
4. What I'd need to see happen to change the answer.
5. Finish with the single sentence I'd tell a friend I decided.
Rules: British English, blunt. Don't upsell me and don't be encouraging. If my answers say stay free, say stay free plainly. Don't quote me a current price, you may be out of date. Tell me to check the site.
Ask me up to three questions before you start.The straight talk
Upgrading won’t make you better at AI. It removes a cap, it adds no skill. A paid plan used timidly is worth less than a free plan used well, and no subscription fixes a vague ask.
Prices move and I can’t see your usage. Treat every number here as roughly right today, and check before deciding.
If your trade needs a specialist tool (design, video, code), that’s a separate question.
Tonight
Audit before you upgrade. Find every AI-ish thing you already pay for, including the wrapper app and the course you never finished, and cancel what you don’t use. Most people fund the upgrade out of the clear-out.
Where this goes next. That’s the money decision made. The Claude Kickstart is the £47 that isn’t a subscription: Claude set up around your trade, five real jobs handed over, one weekend.