Visual Network
Most small businesses I talk to are not short of AI tools. They have two or three subscriptions, a few free accounts, and a nagging feeling that they should be getting more out of them. The problem is rarely the tools. It is that nobody decided what they were for.
1. Start with the job, not the tool
Write down three or four jobs that take time every week. Be specific. Not "marketing", but "writing the monthly newsletter", "answering the same five questions from customers", "turning meeting notes into a proposal". Then ask, for each one: what would a good result look like, and how would I know?
Some of those jobs will not need AI at all. A template, a form or a better spreadsheet may do it. That is a good outcome, not a failure.
2. Check what you already pay for
Before adding anything, list what you have: the AI features inside your email, office software, website builder and accounting package, plus any standalone subscriptions. It is common to find two tools doing the same job and a third nobody has opened since the free trial.
3. Test on your own work
Comparisons online are written about other people's tasks. Take one real job, give the same brief to two or three tools, and compare the results side by side. Score them on what matters to you: accuracy, tone, how much editing they needed, and how long the whole thing took. Keep the scores. You will want them in six months when the tools have changed.
4. Ask where your data goes
For each tool, find out where it processes what you type, whether it may be used to train models, and whether a business plan changes that. Then set one simple rule for everyone: what never goes into an AI tool. Customers' personal details, passwords and anything confidential are a good start. Most of the risk is in everyday habits, and those are easy to fix.
I am not a lawyer, and if you handle sensitive data you should get proper advice. But a one-page rule that people actually follow is worth more than a policy nobody reads.
5. Count the real cost
Per-seat subscriptions add up quietly. Usage-based tools can spike. Work out the monthly cost against the time saved on the jobs from step 1. If a tool saves an hour a month and costs more than an hour of your time, it is a hobby.
6. Give each tool an owner and a review date
Someone should be responsible for each tool: how it is used, whether it still earns its place. Look again every quarter. The tools change fast, and so do the prices.
When a custom tool makes sense
Sometimes none of the general tools fits, because the job depends on your own information or a process only your business has. That is when a small, focused application can be worth building: one job, done well, with a person checking the output. But only after the steps above, so you know exactly what it needs to do.
A second pair of eyes
If you would like someone independent to go through this with you, that is what Consultancy is for. I do not resell tools or earn commission on them, so the answer is whatever fits.