Every week there's a new AI tool promising to save hours, cut costs, and change the way you work. Most of them are fine products solving problems that large enterprises care about. A smaller number are genuinely useful for a business running on a team of five to fifty people. This is a rundown of the tools that actually fit that context — what they do, what they don't do, and what kind of business gets the most out of each one.
For Writing and Content: Claude and ChatGPT (With Caveats)
These two need no introduction, but they need a realistic framing. Claude (from Anthropic) and ChatGPT (from OpenAI) are genuinely useful for business writing: drafting emails, writing website copy, summarising long documents, creating first drafts of proposals or reports. A consultancy that regularly produces client-facing documents can save hours per week using either one as a drafting assistant.
The caveat: they produce plausible text, not necessarily accurate text. For anything factual — especially anything that will be published or sent to clients — you need a human reviewing the output. They also don't know your business. The more context you give them (about your clients, your tone, your specific situation), the more useful they become. A generic prompt gives you generic output.
For small businesses, the practical use cases are email drafting, meeting summaries, repurposing content across formats, and generating first drafts of standard documents. The monthly cost is £15–20 for individual plans. For most business owners, that pays for itself the first week.
For Automation: Make (Formerly Integromat) and Zapier
If you have two or more software tools that should be sharing data but aren't, Make and Zapier are how you connect them without hiring a developer. A restaurant taking bookings via one system and managing customer data in another can use these tools to sync information automatically. A consultancy getting enquiries through a contact form can route them to a CRM, send an acknowledgement email, and notify the relevant team member — all without manual intervention.
Zapier is slightly easier for non-technical users; Make gives you more flexibility for complex workflows. Both have free tiers that cover simple use cases, and paid plans starting around £15–30/month. These are among the highest-ROI tools available to small businesses right now, primarily because the time savings are immediate and the setup is one-time.
The limitation is that they're connectors — they move data between systems that already exist. If your data is in a spreadsheet you maintain manually, or your booking system has no API, there's a ceiling to what they can do.
For Customer Support: Intercom and Tidio
Both handle live chat and AI-powered chatbots, but they serve slightly different business sizes. Tidio is well-suited to small businesses — e-commerce, services, local businesses — that want a chatbot on their website without a large setup investment. It's relatively easy to configure, has a usable free tier, and handles the basics well: FAQ answers, lead capture, simple routing.
Intercom is more powerful and more expensive, typically better suited to businesses with higher support volume and a dedicated team managing it. It has a stronger AI layer that can handle more complex conversations and integrates deeply with CRM and support workflows.
For a small business just starting to automate support, Tidio is the right entry point. For one that's grown into a more complex support operation, Intercom is worth the cost.
For Meeting Notes: Otter.ai and Fireflies
If you spend a meaningful amount of time in client meetings, these tools eliminate the note-taking burden entirely. They join your calls (Zoom, Teams, Google Meet), transcribe the conversation, and produce a summary with action items. A property management company running 10–15 tenant or client calls a week can save an hour or more of note-taking time and get more reliable documentation in the process.
Both tools are around £10–15/month per user. Otter is slightly better for teams that need to share and search transcripts; Fireflies has better CRM integration if you want call summaries pushed directly into your sales pipeline.
The main limitation is that transcription quality drops with background noise, strong accents, or heavy technical jargon. You'll want to review the output, especially for anything going into a client file.
For Document and Contract Work: Notion AI and DocuSign AI
Notion AI has become genuinely useful for small businesses that run their internal knowledge base in Notion — drafting standard operating procedures, summarising long documents, generating templates. It's tightly integrated into the workspace, which means less context-switching.
On the contracts side, AI features in DocuSign and similar platforms (Ironclad, PandaDoc) are starting to add real value: auto-population of standard contract fields, clause identification, and flagging of unusual terms for review. For a law firm or consultancy handling a high volume of standard agreements, this reduces the administrative overhead of contract management without replacing the lawyer's review of anything non-standard.
What to Ignore (For Now)
AI tools pitched at small businesses for predictive analytics, demand forecasting, and customer sentiment analysis are generally not worth the investment at smaller scale. They require data volumes you probably don't have, setup complexity that outweighs the benefit, and they solve problems that are meaningful at enterprise scale but mostly noise at SME scale.
Similarly, AI-generated social media content tools produce generic output that looks AI-generated. Audiences notice, engagement suffers, and the brand damage isn't worth the time saved.
How to Actually Build a Stack
The mistake most small businesses make is adopting too many tools at once, none of them deeply. Pick one problem you want to solve — most commonly, it's time spent on repetitive communication or data entry — and find the right tool for that. Use it properly for 60 days. Measure the time saved. Then decide what to tackle next.
A dental practice might start with Tidio for after-hours appointment enquiries, add Otter for patient consultation notes six weeks later, and evaluate Zapier for syncing their booking system with their email platform after that. Three tools, added over four months, each one earning its place before the next one comes in.
That's how you build a useful AI stack without paying for tools you don't use and without overwhelming your team with too much change at once. The tools are genuinely good now. The question is always whether you've picked the right one for your actual situation.