AI agents are all over the news, but what does that actually mean for your business of five, fifteen or fifty people? Not a robot taking over your company, but software that carries out tasks autonomously: answering questions, looking up data, drafting emails and updating systems. In this article, five applications we see working at SMEs in practice — plus an honest chapter on what AI agents cannot do.
1. Answering customer questions
The most widely used application. An AI agent on your website, in your email or on WhatsApp answers standard questions: opening hours, delivery times, prices, the status of an order, warranty terms. The agent learns from your existing documents, manuals and previously answered emails.
What it delivers: customers get an answer straight away, even in the evenings and at weekends. With webshops we see 60 to 80 percent of incoming questions handled without human intervention. Your team keeps time for the questions that genuinely deserve attention.
2. Following up leads
An enquiry through your website that sits unanswered for three days is a missed job. An AI agent responds within a minute: it thanks the person enquiring, asks a few follow-up questions ("when do you need this?", "how many units are we talking about?") and, if desired, books an appointment straight into your calendar. The lead arrives enriched in your CRM, so you walk into the conversation warm.
What it delivers: more revenue from the same stream of enquiries. Responding faster than your competitor is often already enough to win the job.
3. Drafting quotes
Writing quotes is repetitive work: the same structure, the same texts, different amounts. Based on the enquiry, an AI agent draws up a draft quote with your standard texts, rates and terms. You check it, adjust where needed and send it off.
What it delivers: where a quote used to take an hour, you're now done in ten minutes. With ten quotes per week, that saves a full working day.
4. Summarising documents
Long reports, contracts, minutes, tender documents: an agent reads them and produces a summary with the points that are relevant to you. Ask follow-up questions in plain language: "Does this contract contain any deviating payment terms?" or "What are the three most important requirements in this tender?"
What it delivers: faster decisions without spending evenings reading documents. Mind you: for a legal or financial final assessment, a human eye remains necessary — more on that below.
5. Keeping your calendar and inbox up to date
An agent that scans your inbox, determines urgency, prepares draft replies and schedules or moves appointments. You start the day with a tidy inbox and an overview: this is what you need to do yourself, this is what I have already prepared.
What it delivers: peace of mind. Less switching between tasks, fewer things falling through the cracks. Especially for freelancers and small teams without a secretariat, this is a game changer.
Practical tip: start with one task that costs you the most time each week and follows a predictable pattern. Usually that's answering standard questions. Let the agent run alongside you for two weeks, measure how much time it saves and only then expand. Starting small works better than trying to automate everything at once.
What AI agents cannot do well
Let's be honest: there are limits. An AI agent:
- Doesn't build real relationships. An angry customer with a complex problem wants a human who thinks along and takes responsibility. The agent can prepare the conversation, but not take it over.
- Doesn't know your business automatically. Without good input — your texts, rates, processes — an agent produces average answers. The quality of the input determines the quality of the output.
- Isn't infallible. Agents sometimes make up answers that sound plausible. That's why we always build in control: when in doubt, the agent escalates to a human, and for sensitive documents (contracts, large amounts) human approval is mandatory.
- Doesn't replace craftsmanship. An agent can produce a draft, but the final quality remains your responsibility.
Anyone who tells you an AI agent can run your entire business is selling hype. Anyone who says it adds nothing hasn't seriously tried it. The truth is in between: for recurring, predictable work an agent is astonishingly effective.
What does it cost?
That depends on the complexity, but to give you an idea: we build an agent that answers customer questions based on your existing information from around €2,950 one-off, plus hosting and ongoing development starting at a few dozen euros per month. More complex agents with connections to your CRM, accounting or scheduling run towards €5,000 to €10,000.
Do the maths: do retyping, emailing and quoting currently cost you or your team ten hours per week? Then such an investment often pays for itself within six months. If our calculation says it won't pay for itself, we simply advise against it.
Ready to get started?
Curious which application will deliver the most for your business? Book a free consultation — we'll think along with you for 30 minutes and you'll get an honest estimate of costs and benefits straight away.