When we talk to a business owner about AI, we usually don't start with the technology. We start with the question of where time and enquiries are leaking today. The answers repeat so often that they fit into six points. And for each of them the same thing holds: it isn't a people problem, it's work that should never have been on people.
1. The enquiry arrives at midnight, by morning they're with a competitor
Customers write when they have time, not when you're open. A website form, an Instagram message or an e-mail at ten in the evening waits until the morning, and often longer. Whoever replies first usually wins, and once an enquiry is more than an hour old the chance of a deal drops sharply.
What AI can do: reply within two minutes, 24 hours a day. It confirms receipt, asks two or three follow-up questions, records the enquiry in the CRM and assigns it to a person who opens a ready, sorted list in the morning. The customer doesn't wait and you don't do anything extra.
2. The team answers the same questions
Opening hours, order status, price list, how a return works. Each answer takes a few minutes, but at dozens a day that's hours. Worse, these questions interrupt the work that actually earns the company money.
What AI can do: an agent on your website, in e-mail or on WhatsApp that answers from your own rules and documents, not from the internet. It knows your prices, terms and products. When it isn't sure, it doesn't guess; it summarises the question and hands it to a person with context. In practice this handles most routine questions and the team deals with the complex ones.
3. The same e-mails, forms and reports, every day on repeat
An invoice arrives by e-mail and someone retypes it into the accounting system. An order arrives from the e-shop and someone retypes it into the warehouse. On Friday someone assembles a report from three spreadsheets. This kind of work is tedious, error-prone and nobody enjoys it.
What AI can do: extract the supplier, amount, due date and line items from a document and prepare an entry that a person just approves. Turn an enquiry into a draft quote from a template based on your price list. Send a report every Monday morning that assembled itself. The rule we use: if you do something more than ten times a week and always the same way, it belongs to a computer.
4. CRM here, calendar there, e-mails somewhere else
Most companies have the tools they need. The problem is the tools don't talk to each other. An enquiry from the form is carried into the CRM by hand, an appointment from the CRM into the calendar by hand, the confirmation to the customer is written by hand from e-mail. Every transfer costs time and sometimes a piece of data.
What AI can do: connect what you already have. You don't need to change software. The form, CRM, calendar, invoicing and e-mail can be linked via API or tools like n8n and Make so that data flows on its own. The customer fills in the form, two minutes later they have an appointment in the calendar and you have a new record in the CRM.
5. Follow-ups that never get sent
Every salesperson knows most deals close only after the second or third contact. And every salesperson has a mental list of people they'll "get back to". Usually they don't, because something more urgent is on the desk. An enquiry that never got a follow-up is an enquiry that never turned into a job.
What AI can do: a reminder after three days, an offer after a week, a short message after a month. Personalised to what the customer was dealing with, sent at the right time and stopped the moment they reply. A person decides the content, the system makes sure it actually happens.
6. Nobody knows what's happening in the company right now
How many enquiries came in this week? How many did we close? Which service sells best? When the answers live in five tools and three heads, decisions get made by feel. And decisions by feel are expensive.
What AI can do: collect numbers from the tools you use and show them in one place. Enquiries, conversions, time saved, job status. Not a dashboard for its own sake, but the three to five numbers you actually decide by.
What this means in practice
The companies we've launched automations for describe the same effect: less retyping, faster replies to customers and, finally, an overview. The customer gets a reply at midnight, the invoice books itself, the salesperson opens a list of enquiries in the morning that someone has already sorted. The team doesn't do less, it does the things that require a person.
Where to start
Not with everything at once. Pick the one point from this list that hurts most and measure how much time it takes today. Run the automation for a month alongside the original process and compare. If it saves at least half the time and doesn't cause errors, extend it to the next process. Your data stays in your accounts and the model is sent only what it needs for a decision.
We do a process audit for a fixed price, and its output is a list of what's worth automating, what isn't and how much it saves. Even if we end up building nothing, you'll know where your time is leaking.