How Many AI Tools Exist, and How to Pick One Without Losing Your Mind

Straight answer: there are somewhere north of 20,000 AI tools listed across directories like There's An AI For That, Futurepedia and G2, and that number climbs by the week. Nobody needs to know all of them. What you need is a short, boring filter that tells you within an hour whether a tool earns a place in your business, and I'll give you mine below.

Worth reading next: How to Close Down a Facebook Business Page (Without Losing Data You'll.

So how many AI tools are we talking about

Depending on which directory you trust, the count sits somewhere between 18,000 and 25,000 tools. There's An AI For That passed 20,000 listed tools back in 2024 and has kept growing. Futurepedia lists thousands more, sorted into categories from "AI for writing" to "AI for legal contracts." G2 has an AI-tagged software category with over 1,700 products just in sales and marketing.

Here's the bit that surprises people: most of those tools are not separate products in any meaningful sense. Strip away the branding and a huge chunk of them are the same underlying model, usually OpenAI's GPT-4 or GPT-4o, sometimes Anthropic's Claude, wrapped in a nicer interface with a monthly subscription slapped on top. I've tested "AI email writers" that produce word-for-word identical output to typing the same prompt straight into ChatGPT. The wrapper isn't wrong to exist, some of them save real time with templates and integrations, but the number 20,000 is misleading if you think it means 20,000 different capabilities. It's closer to a few hundred underlying capabilities dressed up 20,000 ways.

Why the sheer number is more of a trap than a benefit

Choice overload is a real cost, not a nice problem to have. I sat down with a client last year, a woman running a 12-person bookkeeping firm in Reading, who showed me her company card statement. Fourteen separate AI subscriptions. Fourteen. A grammar tool, a meeting note-taker, two different chatbot builders, three "AI marketing assistants," an AI invoice reader that did the same job as software she already owned, and so on. Total monthly spend: £412. When I asked her team which ones they opened that week, the honest answer was three.

That's the trap. Every tool markets itself as essential, every landing page has a testimonial claiming it saved someone ten hours a week, and it's hard to tell the difference between a tool that will change how your team works and one that will sit unused after the free trial ends. The fix isn't finding a better directory. It's changing how you decide, before you even go looking.

The uncomfortable part nobody wants to say out loud

Here's the thing most tool round-ups skip: the AI tool almost never fails because the AI was bad. It fails because nobody changed the process around it. I've watched businesses buy a brilliant AI scheduling tool and keep booking meetings by email anyway, because switching habits is harder than switching software. The tool wasn't the problem. The willingness to change how the team works was the problem.

So before you compare five chatbots or three content generators, ask a blunter question: are you willing to change the workflow this tool assumes you'll use? If the answer is no, no tool on earth will fix it, and you'll be back here in six months with a new subscription and the same unsolved habit.

A five-step filter that works

This is the process I run with clients now, and it takes about an hour, not a week of demos.

• Step 1, name the task, not the category. Not "AI for marketing," but "write the first draft of our weekly client email in under ten minutes." Vague categories attract vague tools.
• Step 2, list who does the task now. If it's one person, ten minutes a week, a £20/month tool is overkill. If it's five people doing it daily, the maths for a proper tool changes fast.
• Step 3, work out what a saved hour is worth. A junior admin's time might be £15/hour loaded cost. A senior consultant's might be £80/hour. Multiply that against hours saved per week and you get a real number, not a guess.
• Step 4, test with your own messy data, not the demo data. Every tool looks brilliant on the vendor's clean sample. Feed it your actual invoice, your actual customer email, your actual half-finished spreadsheet.
• Step 5, give it thirty days with one clear kill switch. Decide in advance what "this isn't working" looks like, whether that's a usage number, an error rate, or your own gut after week three, and cancel without guilt if it doesn't clear the bar.

If you want a starting point for step one rather than a blank page, I've broken down which categories of tool map to specific AI tools for different business tasks, which cuts a lot of the browsing time out.

What I use, and what I dropped

I run a small consulting business, so I'll be specific rather than coy about it. I use ChatGPT Plus (£20/month) daily for drafting, research summaries, and thinking out loud. I use Claude for longer documents because it handles bigger chunks of text without losing the thread. I use one scheduling tool that syncs calendars automatically, because chasing meeting times by email was costing me a genuine hour a week.

What I dropped: an AI-powered social media scheduler that promised to write posts for me. The posts sounded like nobody, which is the worst thing content can sound like when your entire business is built on a recognisable voice. I also dropped an AI note-taker for calls, not because it was bad, but because I found I remembered things better when I took my own rough notes, and the transcript became one more thing I never went back to read.

That second one matters more than it sounds. A tool can work well and still not be right for you, because it doesn't fit how your brain operates. Nobody puts that on a features page.

Free tools versus paid, and where the line sits

A lot of businesses overspend early because they assume free means limited and paid means better. Sometimes that's true. Often it isn't. ChatGPT's free tier gives you access to GPT-4o mini with usage caps that reset, and for a huge amount of small business writing and thinking, that's plenty. I've written a full breakdown of exactly how many free ChatGPT questions you really get in 2026, because the caps confuse people more than they should.

The moment to pay is when you hit the cap mid-task more than twice a week, or when you need a feature the free tier locks, like longer context windows, image generation, or the ability to build a custom assistant for repeated tasks. Before that point, paying is just guilt spending because a sales page told you free is for amateurs.

Questions to ask before you sign up for anything

• Does this replace a task I'm currently paying a person or another tool to do, or does it just add a layer on top?
• Can I export my data out cleanly if I cancel, or am I locked in?
• Does the free trial let me test with my real files, or only their sample content?
• Who on my team will be responsible for using this daily, by name?
• What's the cancellation process, and is it a two-click cancel or a "call to cancel" trap?

That last one weeds out more bad tools than any feature comparison. If a company makes cancelling difficult, that tells you something about how confident they are that you'll want to stay once the excitement wears off.

Where AI tool choice overlaps with other tool decisions

The logic here isn't unique to AI, and that's worth saying plainly. I've written about choosing the right time tracking tool using almost the exact same steps: name the task, count the real users, test with messy real data, give it a fixed trial with a kill switch. The AI label makes tools feel newer and scarier to evaluate than they are. They're still software. They still need to earn their monthly fee against a real task.

If your team already has policies about which software gets approved for use, this is also the moment to check whether AI tools fit inside them or need their own rules, particularly around data handling. I've covered the wider question of whether AI tools should be allowed in workplaces at all and under what conditions, which is worth a read before you roll something out team-wide rather than just for yourself.

For mobile, here are AI apps for your phone.

When to stop choosing tools yourself and get help

If you've read this far and still feel stuck between six shortlisted tools, that's a sign the problem isn't the tools, it's that nobody in the business has mapped which tasks need automating in what order. That's a half-day piece of work with the right person, not a month of trial subscriptions. I do this with clients directly, and if you want a sense of what that looks like and costs, I've laid it out on my AI consultant for small business page rather than making you guess at a quote.

A good place to start narrowing down is this list of ChatGPT alternatives.

The short version to keep on hand

Ignore the total count of tools out there. It's noise. Pick one task, put a real cost on the time it eats, test the tool against your ugliest real data, give it thirty days with a clear cancel rule, and only then decide if it earns a permanent line on your budget. Most of the 20,000 tools out there will never touch your business, and that's completely fine.

A closely related page: AI write for us.

Related reading: how many accounts are on tiktok.

Related reading: cuantos canales hay en youtube.

For the bigger picture on AI strategy for your business, see my guide.

Frequently asked questions

How many AI tools exist right now, in 2026?

Directories like There's An AI For That and Futurepedia list somewhere between 20,000 and 25,000 AI tools, though many are thin wrappers around the same handful of underlying models, so the real number of distinct capabilities is far smaller.

What's the biggest mistake businesses make when choosing an AI tool?

Buying based on the demo rather than testing with their own messy real data, and never setting a clear point at which they'll cancel if it isn't pulling its weight after thirty days.

Should a small business use free AI tools or pay straight away?

Start free wherever a free tier exists. Only move to paid once you hit usage limits more than twice a week or need a specific feature the free version locks, like longer documents or a custom assistant.

Is it worth hiring someone to help choose AI tools instead of doing it myself?

If you're stuck comparing several similar tools without a clear task, cost, and success measure for each one, a short session with someone who does this daily usually saves far more than the subscription cost of guessing wrong three times in a row.