How Many AI Tools Are There Like the Big Chatbots? I Went and Counted

Straight answer: Directories list somewhere between 20,000 and 30,000 AI tools right now, but only around seven or eight companies build the underlying models that power nearly all of them. Almost everything else you'll find advertised as "the new ChatGPT" is a wrapper, a fine-tune, or a rebrand sitting on someone else's technology. That's the number that matters when you're deciding what to use for your business, not the eye-watering directory total.

The 20,000 figure, and why it's basically meaningless

There's An AI For That, one of the biggest AI directories online, has listed over 20,000 tools since sometime in 2025. Futurepedia sits somewhere around 8,000 to 10,000 depending on the week you check. Add smaller directories like AI Tool Hunt, Toolify and a dozen Product Hunt collections, and you could tell a client "there are 30,000 AI tools out there" and not be lying.

But that number counts things you'd never touch. It counts abandoned side projects a developer built over a weekend and never updated again. It counts three separate listings for the same Chrome extension under three different names. It counts tools that shut down in early 2025 but nobody removed the listing. When I went through 40 "top chatbot alternatives" listed on one popular directory for a client last year, 11 of them returned a 404 page, 6 hadn't been updated since 2024, and only 3 had a pricing page that worked.

So the honest answer to "how many AI tools are there like the big chatbots" has two layers: a huge, mostly meaningless directory number, and a much smaller, much more useful number underneath it.

Underneath it all, there are about eight companies

Here's the bit that most listicles skip over, because it makes the "20,000 tools" headline less exciting. Almost every chatbot you'll ever be pitched runs on one of these foundation models:

• OpenAI, which powers ChatGPT and is licensed out through its API to hundreds of other branded chatbots
• Anthropic, which builds Claude and licenses it the same way
• Google, with Gemini, built into Workspace, Search and countless third-party apps
• Meta, with the open-weight Llama models, which anyone can download and build a chatbot on top of for free
• Mistral, a French company whose smaller models turn up inside a surprising number of European "sovereign AI" tools
• xAI, with Grok, mostly seen inside X and a growing number of standalone apps
• DeepSeek and Alibaba's Qwen, both Chinese, both open-weight, both increasingly common inside cheaper Western-branded tools that never mention where the intelligence comes from

Stanford's AI Index report, one of the more careful pieces of research on this whole industry, has tracked this same pattern for several years running: a small number of labs produce the frontier models, and a huge, fast-moving layer of startups builds interfaces, prompts and workflows on top of them. You can read the full report on Stanford HAI's AI Index site if you want the underlying data rather than my summary of it.

What this means practically: when you see a new "AI chatbot" branded as some startup's own invention, the honest question isn't "is this a good tool", it's "whose model is this running on, and what am I paying extra for." Sometimes the answer is a useful workflow layered on top. Often the answer is nothing much beyond a nicer login screen.

A real example from a client review

Last year a bookkeeping firm near Guildford asked me to review a shortlist of 12 "AI chatbot assistants" they'd found through a LinkedIn post promising to save them ten hours a week on client emails. I ran each one through the same four questions: what model is underneath it, what happens to client data once it's typed in, what does it cost at the volume they'd use it, and what happens if the company folds.

Of the 12, nine were running on OpenAI's API with a different logo on top. Two were built on Claude. One, a UK-based tool aimed specifically at accountants, had built its own retrieval layer on top of an open-weight model and was doing something different, mostly around how it handled document uploads and kept them separate from the model provider's own training data. That one was worth the money. The other 11 were charging between £15 and £45 a month for essentially the same underlying capability the firm already had access to through a £20-a-month ChatGPT Plus subscription, just wrapped in accounting-flavoured branding.

This is the part that most "best AI tools" roundups won't tell you, because plenty of them earn affiliate commission on every tool they recommend, and a longer list means more commission opportunities. The uncomfortable truth is that the AI tools market right now rewards volume, not originality. Building a chatbot wrapper takes a developer a weekend and an API key. Building something new on top of one takes months, real product thinking, and usually a founder who understands the industry they're building for. Both get listed as "AI tools" in the same directory, at the same size, with the same star rating.

What separates the big chatbots from each other

If you strip out the wrappers and just compare the handful of models that matter, the real differences are smaller and more practical than the marketing suggests:

• ChatGPT (OpenAI) has the biggest plugin and integration ecosystem, the most third-party tools built on it, and at £20 a month for Plus is the default most people already know
• Claude (Anthropic) tends to write more naturally for long-form content and handles longer documents well, also around £20 a month for Pro
• Gemini (Google) is the one to use if your business already lives inside Google Workspace, since it sits directly in Docs, Gmail and Sheets, priced around £18.99 a month through Google One AI Premium
• Perplexity behaves more like a research assistant than a chatbot, citing sources as it goes, also roughly £20 a month for Pro

None of these differences are dramatic enough to justify most businesses paying for more than one or two. The bigger decision isn't which of the big four to pick, it's whether you need a general chatbot at all or something more specific, like a tool that reads your invoices or drafts your social posts inside your existing workflow. If you're still fuzzy on what counts as "an AI tool" versus a chatbot versus something else entirely, I wrote a plain-English breakdown of what AI tools means for exactly this confusion.

Step by step: how to shortlist from 20,000 down to two

This is the process I use with clients when they're drowning in options, and it takes about 90 minutes:

• Step 1: write down the actual task, not the tool category. Not "I need an AI chatbot" but "I need something to draft first replies to customer emails in under 30 seconds"
• Step 2: check whether one of the big four (ChatGPT, Claude, Gemini, Perplexity) already does this well enough on its own, because if it does, you don't need tool number 20,001
• Step 3: if you need something specific to your industry, find out which underlying model it's built on, and ask the vendor directly if they won't say
• Step 4: check the data handling. In the UK, this matters more than most business owners realise, and I've written a full comparison of which AI tools are GDPR compliant versus which just say they are because the gap between the two is bigger than you'd expect
• Step 5: test it on a real task from your business for a week before paying annually, since monthly plans exist for exactly this reason
• Step 6: ask what happens to your setup, your prompts and your data if the company shuts down or gets acquired, because a huge chunk of this market will not exist in three years

If step 3 feels like too much digging around for one person to do alongside running a business, that's fair. This is more or less what happens in the first session when someone works with an AI implementation coach, mapping out what you already have against what you need before you spend another penny on tool number 47.

Where the count is heading next

Two things are pushing that 20,000-plus number higher every month, and it's worth understanding both if you want to stop feeling like you're behind.

The first is agents, which are a step beyond chatbots because they can take actions rather than just answering questions, booking things, sending things, checking things without you prompting each step. I did a proper count of this specific category separately, because it's exploding even faster than plain chatbots right now, in a piece where I counted how many AI agents are worth trying. Spoiler: the useful number there is even smaller than it is for chatbots.

The second is price. A striking number of these tools are free, or free enough to feel free, and that's not generosity. It's a customer acquisition strategy, usually funded by venture money that expects a chunk of free users to convert to paid tiers or to have their usage data feed the next model version. I've broken down exactly why so many AI tools are free and how they're making their money, and it changes how you should think about trusting a free tool with sensitive client information.

One more thing worth knowing if you're picking between chatbot tools for anything customer-facing: the personality and boundaries of a chatbot aren't magic, they're written instructions sitting behind the scenes called system prompts, and understanding roughly how they work will change what you expect from any tool you buy. I cover this in what system prompts are and why they matter for AI chatbots.

The number that should guide your decision

So here's where I land after doing this exercise for dozens of clients over the past two years. There are roughly 20,000 to 30,000 AI tools listed across the internet's various directories. There are around eight companies building the actual technology underneath nearly all of them. And for most small businesses, there are usually only two or three tools worth paying for at any one time, one general chatbot for thinking and drafting, and maybe one specific tool built for the one repetitive task that's eating the most hours in your week.

Everything else is noise, and noise is exhausting to keep up with. You don't need to try 40 chatbots to know which one works for your business. You need to try three, for a week each, against a real task, and pick the one that saves you time rather than the one with the most persuasive landing page.

Auf Deutsch: Strategien zur Kostenreduzierung im Marketing-Tech-Stack.

See also the AI guest post page.

Frequently asked questions

How many AI chatbot tools are there in total?

Directories list somewhere between 20,000 and 30,000 AI tools overall, but the vast majority are wrappers or interfaces built on top of a small number of underlying models from companies like OpenAI, Anthropic, Google, Meta and Mistral, rather than separate technologies.

Is ChatGPT still the biggest AI chatbot?

Yes, in terms of user numbers and the size of the ecosystem built on its API, ChatGPT remains the most widely used chatbot, though Claude, Gemini and Perplexity each do specific things better depending on your task, particularly long-document work, Workspace integration and cited research respectively.

Do I need more than one AI chatbot for my business?

Most small businesses only need one general chatbot for drafting and thinking, plus at most one specific tool built for a single repetitive task, such as invoice reading or social post drafting, rather than a stack of five or six overlapping subscriptions.

Why do so many AI tools look identical to each other?

Because most of them are built on the same handful of underlying models through publicly available APIs, meaning the interface and branding can differ enormously while the actual intelligence generating the answers is coming from the same one or two sources across dozens of competing tools.