AI that earns its place in your business
Most small businesses don’t need an AI strategy. They need the quote written, the invoice chased and the customer replied to before lunch. We publish tested workflows for tools you can actually afford, and we tell you when AI is the wrong answer.
Independent · No paid placements · Every guide reviewed before publishing
What to hand over, and what to keep
The failure mode with AI is not bad writing. It is confident, plausible nonsense sent to a customer. Draw the line before you start, not after.
Worth handing over now
- First drafts of routine customer replies, which you then edit
- The structure of a quote or estimate, before you set the numbers
- Turning a rambling client call into a written summary
- Service and product descriptions you keep rewriting
- Converting a job you already know into a repeatable checklist
Keep in human hands
- Final prices, discount rules and anything margin-sensitive
- Contract terms and legal wording
- Apologies and complaint responses
- Customer data you have not cleared for that tool
- Any decision where money is committed without review
Tools small businesses actually run on
Sixteen tools that show up in real one and two person businesses, with what each is genuinely good at rather than what its landing page claims.
ChatGPT
OpenAI · Assistant
General drafting, planning and customer replies in one place.
Best for: Trying AI for the first time
Claude
Anthropic · Assistant
Strong on long documents, rewrites and careful editing.
Best for: Turning rough notes into client-ready work
Gemini
Google · Assistant
AI built into Gmail, Docs and Drive where the work already sits.
Best for: Businesses already living in Google Workspace
Microsoft Copilot
Microsoft · Assistant
AI inside Word, Excel, Outlook and Teams.
Best for: Businesses standardised on Microsoft 365
Canva
Canva · Design
Templates and AI layout tools for quotes, social posts and one-page flyers.
Best for: Marketing without a designer
Grammarly
Grammarly · Writing
Catches tone and clarity problems before an email goes out.
Best for: Fast, polished client email
No paid placements. Prices and free tiers change, so check the tool's own pricing page before you commit.
Two guides that pay for themselves first
Both replace a job you already do every week, and both end with a template or checklist you keep.
How to write quotes and estimates with AI
Build a quote from a rough scope, then check it for the three things that quietly eat your margin.
Customer replies that do not sound like a robot
Turn the twenty questions you answer all day into a reply system with your own voice and an escalation rule.
Free plan or paid: what actually changes
Tool pricing pages sell features. For a small business, five things decide whether the free tier is enough. This is the pattern across the major assistants, automation platforms and design tools, not a claim about any single price list.
| What changes | Free tier, typically | Paid tier, typically |
|---|---|---|
| Usage limits | Daily message or task caps, throttled when the service is busy | Higher or pooled limits, priority at peak times |
| Your data | Inputs are often used to improve the model | Usually excluded, under a data processing agreement |
| Team access | One login, one person | Shared workspace with roles and admin controls |
| Connections | Manual copy and paste between apps | Connectors to your inbox, CRM, calendar or invoicing |
| Support | Help centre and community | Email support with a response commitment |
The rule we apply: stay on free until a task repeats every week, then pay for the data terms before paying for the extra limits. A tool that trains on your customer emails is more expensive than it looks.
Pick the task, not the tool
Write down the job you do every week and resent, then find the guide that removes it. Starting with a tool means paying for something you never open.
How these guides are tested and kept current
We run the job, not the demo
Every workflow is carried out in the tool’s current interface, including the awkward parts: uploading a real document, fixing the formatting, and seeing what happens when the output is unusable.
We label example output
Sample quotes, emails and summaries on this site are illustrative unless stated otherwise. No invented case studies, no borrowed customer stories, no screenshots passed off as our own results.
We date it and correct it
Each guide carries its last review date. When a tool changes its interface, limits or data terms, the guide is rechecked and either updated or the outdated claim is removed.
Full standards, including how to report an error: editorial policy.
Questions owners ask before they start
Do I have to pay for a tool to start?
No. Every workflow here can be attempted on a free plan. Paid tiers start to matter when you need team seats, higher usage limits, or better handling of customer data. Where a step depends on a paid feature, the guide says so.
Is it safe to paste customer details in?
Not by default. Consumer free tiers often use your input to improve their models, while business tiers usually contract out of that. Strip names, addresses and account numbers, or use a business plan with a data processing agreement. If you would not put it on a postcard, do not put it in a free chatbot.
Will AI-assisted content hurt my Google ranking?
Google’s guidance targets unhelpful, mass-produced pages rather than the tool used to draft them. A page that answers a question better than the competing results is fine. Automatically publishing thousands of thin pages is what gets sites demoted.
How long before this actually saves time?
Expect the first week to be slower than doing it by hand, because you are writing the instructions you will reuse. The saving appears when a task repeats weekly and the prompt is already written down.
More on how these guides are produced: our editorial policy.
The six words that confuse everyone first
You do not need the vocabulary to use these tools, but six terms come up in every guide. Here is what they actually mean.
- Prompt
- The instruction you give the tool. A specific brief with your rates, tone and exclusions produces usable output; a vague question produces generic text.
- Hallucination
- A confident answer that is simply wrong, stated without hesitation. The reason you check every number, name and quotation before it reaches a customer.
- Context window
- How much text the tool can hold in mind at once. Long contracts and spreadsheets get truncated, and what falls out of the window cannot be reasoned about.
- Agent
- A tool that takes actions rather than drafting text: sending, booking, filing, updating a record. More useful and more dangerous, because mistakes propagate without you in the loop.
- Fine-tuning
- Training a model further on your own examples. Rarely worth it before you have good prompts; most businesses never need it.
- Token
- The unit models are billed in, roughly three quarters of a word. This is why limits are quoted in tokens rather than pages.
Browse by topic
Five areas, in the order most small businesses need them.
