test
How much don’t you know about AI?
7 questions, 2 minutes, no email. It isn’t an exam: it’s a mirror — every answer shows you something most companies find out late.
Question 1 of 7
How many hours a week does your team lose to repetitive tasks (moving data around, answering the same thing, copying from one place to another)?
You’re the exception — genuinely. The sharper question is whether the automation stopped too early: there’s almost always a second layer of hours nobody has looked at.
“Normal” is the key word: most companies have normalised hours a system could already take on. Nobody counts what they take for granted — and so nobody gets it back.
Most small and medium companies never count them. Putting a price on them usually stings — and that is exactly the first step to getting them back.
If ChatGPT and the rest disappeared tomorrow, what would change in your company?
You may not use them, but your competitors do. And ChatGPT is only the front door: there are tools that could already be taking work off your team every week.
Everyone doing it their own way is the most common state of play. It’s also the most expensive: same cost, no method — and the knowledge walks out when that person does.
Method is the word almost nobody gets to use. On that footing, the next jump — automating properly — is much shorter than it looks.
Is there any rule in your company about what data can be pasted into a public AI?
Rare and valuable: very few companies have it in writing. That one rule protects more than plenty of expensive security tools.
Instinct doesn’t scale: every person decides differently, every day. Anything pasted into a public AI leaves the company — clients, prices, contracts.
Right now someone on the team could be pasting client data into a public AI without knowing it leaves the company. That’s not on them: nobody has written the rule.
How many tools do you pay for every month and only half use?
Not common: spending under control. The refined move from here is building your own tool only where it genuinely pays — and then it’s yours for good.
The most common silent leak: half-used subscriptions nobody reviews. Sometimes building your own works out cheaper — no monthly fee, and made for your case.
It happens in most companies: the fees pile up and nobody reviews them. Just listing them already saves money; replacing the ones you can do without saves rather more.
Would anyone on your team be able to tell an answer the AI made up from a true one?
That is THE skill. With it, everything else — processes, automation, tools — stands on solid ground instead of on blind trust.
That’s where the risk sits: AI makes things up with complete confidence, and it only takes one trusting person for a mistake to reach a client with your name on it.
AI invents facts with absolute confidence — it’s called “hallucination” and it comes with no warning. The good news: checking is a skill you can teach in hours, not months.
A client writes to you at 11pm on a Saturday. What happens?
Two days of silence is an eternity for someone comparing options. Replying — or at least sorting the message — out of hours no longer needs anyone awake.
It works… at somebody’s expense. A system that replies and sorts gives the weekend back — without losing the client who won’t wait until Monday.
That puts you ahead of most. There is a level above: having the reply drafted and waiting first thing on Monday.
How much do you think it costs to properly get started with AI in a company like yours?
That’s the myth that stops most people. You can start small, measurable and with a clear return — what’s genuinely expensive is starting late.
Fair enough: every case is different and nobody publishes prices for this. That’s why you start small and measure — without the huge commitment people tend to imagine.
Exactly. And whoever knows that is usually one step from doing it: the first small process, well chosen, pays for the next ones.
result
[In the dark.]
- There’s no AI working inside the company yet — and that’s where most start, not a problem.
- The first thing isn’t a tool: it’s knowing where it applies and where it doesn’t pay off.
- Two clear use cases get you further than ten scattered experiments.
result
[Everyone their own way.]
- AI is already inside the company, but everyone uses it however they like.
- The jump isn’t a new tool: it’s a shared method and clear rules of use.
- Training the team on real cases turns scattered use into repeatable work.
result
[Leaking.]
- The basics are there: there’s use, there’s judgement and there’s some method.
- What’s left are the leaks — hours that repeat every week, or tools you pay for and half use.
- You plug them one at a time, starting with the one that costs most.
result
[Leaking.]
- The basics are there: there’s use, there’s judgement and there’s some method.
- What’s left are the leaks — hours that repeat every week, or tools you pay for and half use.
- You plug them one at a time, starting with the one that costs most.
result
[Dialled in.]
- There’s little we can tell you that you don’t know: there’s use, there’s method and there’s judgement.
- At this level what’s usually missing isn’t training, it’s a product of your own.
- If you want to go further, let’s talk about custom software.