The short answer
Start with a general language model for the whole team (ChatGPT Business or Microsoft 365 Copilot). Then add an automation tool (Make or n8n) for work that comes back every week. Only after that come specialised tools for administration, customer contact or marketing. Buying a tool before the process is clear costs money and delivers nothing.
The order that works
| Order | Category | What to think of | Indication |
|---|---|---|---|
| 1 | General language model for everyone | ChatGPT Business, Microsoft 365 Copilot, Claude | 20 to 30 dollars per user |
| 2 | Automation of recurring work | Make, n8n, Zapier | 9 to 20 dollars, or free when self-hosted |
| 3 | Meeting notes and transcription | The built-in features of your meeting tool | Often already included |
| 4 | Specialised per discipline | What your accounting, CRM or marketing package offers itself | Varies, check what you already have first |
Step 1: a language model for the whole team
A licence for one or two people delivers nothing. The effect only appears once everyone who writes, searches or researches uses it. Choose between ChatGPT and Copilot based on where your documents live, and make sure it is a business licence: then there is a data processing agreement and no training on your data.
Step 2: automation for what comes back every week
This is where the money starts. A chat window waits until someone asks something; a workflow runs by itself. For most small and medium businesses Make is the wisest starting point; at high volume or with data that has to stay on your own servers, n8n is. See the comparison.
What you can skip
- A separate AI chatbot on your website, as long as your incoming mail is still sorted by hand. The problem sits further down the line.
- Tools promising to take over your entire administration. They take over the entry, the checking stays with a person.
- A second language model because it scores slightly better on a leaderboard. The difference in use is smaller than the difference in how well you apply it.
- Anything with AI added to the name whose function your existing package already has.
Order matters more than choice
Most failed AI rollouts I see did not fail on tool choice but on order: a tool bought first, a problem sought afterwards. Start with the process that comes back most often. The free process check gives you fit, unsure or no-go in seven questions.
How this list was made
This list does not look for the most tools, but for what a small or medium business earns back first in each category. The list follows four questions: what does it cost per user per month, does a business licence come with a data processing agreement and training on your data switched off, does your existing software already have the feature, and can an employee check the outcome before anything happens.
Prices and features come from the official vendor websites, checked on 14 September 2026. They change, so check with the vendor before you buy.
Frequently asked questions
What is the best AI tool for a small company?
There is none. For almost every company the first meaningful step is a business licence on a general language model for the whole team, followed by an automation tool for work that comes back every week.
How much should I budget per month?
For a team of five, a language model plus an automation tool comes to roughly 120 to 180 euros per month in licences. The biggest cost is not the licence but the time to set it up properly.
Do I need specialised AI software?
Usually only as step three or four, and often not: many accounting, CRM and marketing packages already have the feature built in. Check what you already have before buying anything.
Which tool automates my inbox?
No single tool does that by itself. It is a workflow: sorting, collecting data, flagging missing information and preparing a draft that an employee checks and sends.
About the author
I am Joost, AI specialist and founder of Tamboer AI. I have worked with AI daily since 2022. I tested and developed applications at one of the largest banks in the Netherlands and now build AI workflows for smaller businesses, with a person who checks every outcome. More about me.
