There are now more AI tools than there are hours in which to evaluate them, which is itself a problem worth naming.
So this is not a list of fifty. It is six categories that matter to a small business, what I pay for in each, and what I tried and dropped.
Two caveats before we start. Prices and features move monthly in this market, so check before you buy. And the single most common mistake I see is buying a specialised tool for something the general assistant you already pay for does adequately.
That second point will save you more money than anything else in this article.
Quick Info
The Rule I Buy By
Before any tool, one test. Does this replace a task I do repeatedly, and can I describe that task in a sentence?
If yes, it is a candidate. If the answer is that it seems useful and everyone is talking about it, it is a subscription you will forget to cancel.
Buy tools for tasks, not for capabilities.
On that basis, most small businesses need one general assistant and two or three specialists. Not fourteen.
One: The General Assistant
This is the one non-negotiable purchase, and it does more of the work than every specialised tool combined.
The three serious options are ChatGPT, Claude and Gemini. They are close enough in capability that the differences that matter are practical rather than technical.
Pick one, pay for the proper tier, and learn it properly. Running three at once mostly produces three half-learned tools.
The paid tier is worth it. The free tiers are demonstrations.
Two: Meeting Notes and Transcription
The clearest return of any category, because the task is unambiguous and nobody wants to do it.
Any of the main note-takers will join a call, transcribe it and produce a summary with actions. The differences are in accuracy on accents and jargon, and in whether it integrates with your calendar and CRM without effort.
What I would judge on: does it handle your accent and your industry vocabulary, does it join automatically, and can you search across every past call. That last feature is the one that quietly becomes indispensable.
One warning. Tell people they are being transcribed. In several jurisdictions you are legally required to, and in all of them it is the decent thing.
Three: Writing and Content
This is the category most oversupplied with tools you do not need.
A general assistant with a good prompt does the vast majority of content work. The specialised writing platforms mostly add templates, a brand voice setting and a workflow, and whether that is worth a second subscription depends entirely on whether several people need to produce content consistently.
Solo operator: skip them. Team of five producing content weekly: possibly worth it for the consistency.
What is genuinely worth paying for separately is a good editing tool for grammar and clarity, because it catches things you will not, and it costs very little. I go into the writing side properly in my writing tools comparison.
Four: Design and Visuals
Two separate needs here, and people conflate them.
For layout work, social posts, presentations and documents, a template-based design platform with AI features built in will serve almost every small business. The AI features are secondary; the templates and the brand kit are the value.
For image generation, the models are impressive and the output is still recognisable as generated, which matters for brand work. I use it for internal material, mockups and illustrative images, and I use real photographs for anything customer-facing.
My honest advice on generated imagery for a small business: a decent phone photograph of your actual work beats a generated image every time, for credibility and for local SEO.
Five: Automation
The highest-leverage category once you have a process worth automating, and the easiest to waste time on before then.
The pattern that works is a workflow tool connecting the apps you already use, with an AI step in the middle doing the judgement part. Read the enquiry, classify it, draft a reply, file it, notify someone.
Start with one workflow. The one you do most often and resent most. Build it, run it for a month, then build a second.
What not to do is buy the platform and then look for things to automate. That produces eleven half-built workflows and no time saved. My automation article has the specific ten I would start with.
Six: Customer Communication
AI-assisted inbox and chat handling is now genuinely good, with one important boundary.
It works well for drafting replies a human then sends, for classifying and routing enquiries, and for answering factual questions from your own documentation.
It works badly, and visibly, as an unsupervised front line for anything emotional, complex or commercially sensitive. Customers can tell, and the failure mode is a lost customer rather than a slow reply.
So the setting I recommend is draft, do not send. The time saving is nearly the same and the risk is a fraction.
What I Tried and Dropped
Worth including, since every list of this kind pretends everything worked.
The common thread is that they all automated a task that was not actually the bottleneck.
A Realistic Stack and Budget
For a solo business or a small team, this is what I would actually buy.
That comes to a modest monthly total and covers the great majority of what AI can usefully do for a small business today.
Everything else can wait until you have a specific task it removes.
Frequently Asked Questions
A general assistant at the paid tier. It covers more ground than any specialised tool and teaches you what you actually need next.
No. Running several usually means learning none of them properly. Pick the one that fits your working style and go deep.
For a solo operator, rarely. For a team needing consistent output from several people, sometimes. Test against your general assistant with a good prompt first.
For internal material and mockups, yes. For customer-facing brand and local content, real photographs of your actual work perform better and read as more credible.
Social media autopilot tools. Automated scheduling is useful. Automated writing of posts reads exactly as automated.
Enough for one general assistant plus two or three specialists tied to specific tasks. Beyond that you are usually buying capability you have no process for.
Before You Go
The pattern across everything above is that the tool is rarely the constraint. The process is.
Buy one assistant, learn it properly, then add a specialist only when you can name the task it removes.
The full current list of what I use, with the specific configurations, is on my tools page, and there is more in the resources section.
If you want help working out which three tools your business actually needs, get in touch.
One assistant, learned properly, Tariq
I'm a marketing consultant, entrepreneur and content creator. I help businesses grow through practical marketing, websites, SEO, content and AI.
More About Tariq →