AI-implementatie

12 concrete AI applications for SMEs

Marc Cornelius2 November 2026 · 4 min read

AI for SMEs on LinkedIn is mostly about trendy tools you have to install this week. That teaches you nothing. The question is not which tool is newest, but which task you can hand off today. Below are twelve applications we see working in practice, with four real cases attached. And each time the question: where can the machine carry on, and where does a human need to stay involved?

1 to 4: the work nobody enjoys

Start with the boring stuff. Drafting quotes from an email or a message. Summarising documents and long threads down to a decision. Reports that are now plucked by hand from five different tools. And the classic retyping: keying data from one system into another. This is repetitive, it costs highly paid people hours, and nobody enjoys doing it. Which is exactly why it is the first thing you automate.

5 to 8: sales and marketing

Gathering prospect lists and market data that would otherwise take days. A first draft of a newsletter or blog that a human then sharpens. On-brand imagery for social and mockups, without needing a designer. And sorting and enriching incoming requests so your sales people only speak to the good leads. AI does the groundwork, the human adds the finishing touch and the contact.

9 to 12: systems and access

A custom planning tool that removes the sloppy Excel errors. A digital signing alternative so quotes are signed in seconds instead of print, sign, scan, email. Client access that runs through a tidy link rather than manual per-platform explanations. And as the capstone: separate automations coming together into one system, a Business OS where your whole business runs in one place. From loose hacks to something that connects.

Adverge: 80 percent shorter hiring time

For Adverge we automated candidate testing. Applicants show their skills in a simulated Google Ads account, without access to real client data or real budget. Managers then only speak to people who have already passed the simulation. Hiring time dropped by 80 percent, it saves the managers over fifteen hours a week, and the system screens a thousand candidates as easily as ten.

Argo: taking out the retyping

At Argo the pain was not in the expertise but in the retyping. Dimensions in Excel, formatting in Word, print, sign, scan, email. For every quote, over and over. We built a quote generator that removes that manual work and all that printing and scanning. The specialists get back to what they are good at, the machine handles the admin around it.

“The question is not which tool is newest, but which task you can hand off today.”

TOADS: hundreds of locations out of the sheets

TOADS tracked nearly seven hundred locations across five brands and two countries in separate spreadsheets. Adding new participants, connecting data sources and checking who still had no business profile or analytics: error-prone manual work. We built a custom CRM for it with all sources linked to the location, controllable from Claude via a dedicated connection. Now the whole network sits in one system instead of a pile of sheets.

LinkMyAgency: from 30 minutes to 30 seconds

LinkMyAgency is our own SaaS. Where an agency used to explain by hand, per platform and per client, how to grant access, the client now handles it with a single branded link. Client access went from thirty minutes to thirty seconds per onboarding. More than a hundred agencies now run on it. Connecting goes through OAuth, so no passwords are shared or stored.

Where the human is indispensable

In all these applications a human sits at the wheel somewhere. AI tends to be a bit too eager and has no taste. We once built a quote tool for a box supplier. In came a request with a box type and a layout that made no sense at all. The AI approved it straight away. A staff member followed up anyway to check it was right. Turned out to be a mistake. That is humans-in-the-loop: the machine does the volume, the human catches the blunder.

Which hype I would avoid

Not everything that can be done, should be. Never put AI on the part where your customer expects a human: cold outreach, first contact, first impressions. And never let tools with sensitive client data be simply clicked together without a developer finishing the security. The hype says: automate everything. I say: automate the repetitive, keep the human on the parts that matter.

Where would you start?

Twelve applications, and yet you never start with the list. You start with the question of which half hour is the most expensive and most painful for you. Want to know which of these applications would pay off fastest at your company? Book an introduction. Then we walk through your processes together and pick out the best starting point.