co-Founded by AI
NL Subscribe
Live experiment
· Geert-Jan & Jurriën
Week 22

Future of AI in SMEs

SMEs, put your AI agents to work

The Dutch entrepreneur faces a problem: talent remains scarce, costs keep climbing, and productivity growth lags. Meanwhile, a technology sits on the table that fundamentally transforms much of our daily work. Yet many companies still treat AI as a handy text writer. That's roughly like buying a truck just to turn on the radio.

Three recent contributions sharpen the debate. VNO-NCW chairman Coen van Oostrom writes in the FD that our future prosperity depends on working smarter with the same people. With an aging labor market, productivity gains are no luxury. Investor Robert Verwaayen goes further. He sketches an SME where a few entrepreneurs collaborate with an army of AI agents. And entrepreneur Thijs Verheul shares his own discovery journey with Claude Code and Codex on LinkedIn. His message to peers: stop watching, start building.

I recognize their enthusiasm. At co-Founded by AI, Jurriën Kerstholt and I experiment every day with that exact thinking. We're building a product company with one physical product, no employees. Our goal: €200K revenue in year one, with a maximum of four hours per founder per week. AI ideally handles 95 percent of execution work. We make the decisions.

It's not a proven formula. It's an experiment with a fairly sharp yardstick. And precisely because of that, we learn so much. AI helps us with market research, product development, supplier selection, copywriting, website, and customer contact. What once required different specialists and agencies, we're organizing step by step with agents. One agent researches, another executes, a third reviews. We set direction, establish boundaries, and assess outcomes.

That sounds more orderly than reality. An agent can produce a convincing analysis based on flawed assumptions. With too much access, it can also reach data outside its scope. That's why we build a shared corporate memory alongside the agents, clear work instructions, permissions, and oversight. A virtual employee needs a job description and a manager just like anyone else. The digital HR department can skip the holiday party for now.

For SMEs, this is where the real opportunity lies. A large company has budgets, IT departments, and innovation teams. A small entrepreneur has something else: speed. You can decide Monday morning to restructure a process and run the first experiment Tuesday. Without a steering committee of twelve.

Start with the daily frustration. Quotes sitting on the shelf. Customer questions that keep recurring. Inventory running out at the wrong time. Reports a staff member spends hours on every Friday. Give an agent a defined task, access to the right data, and clear boundaries. Let it observe first and make suggestions. Only then give it space to act independently. Measure how much time it saves, how many errors emerge, and whether customers are served better.

The goal isn't a company where people disappear. The accountant, buyer, or customer service rep gains space for complex cases, relationships, and better decisions. AI strips routine from their work and gives small teams the firepower of larger organizations. That demands more than a subscription to a language model. It demands curiosity from the entrepreneur themselves.

Waiting for a ready-made solution is tempting. Only your competitor is meanwhile learning every week. For many SMEs, this shift is about more than efficiency. It affects their competitive edge and ultimately their survival.

So my call to Dutch SMEs: don't make AI a sidecar project for the IT person or intern. Put it on the entrepreneur's agenda. Experiment with agents, discuss the failures and share the wins. We do that at co-Founded by AI daily, with all the wonder and frustration that comes with it.

SMEs, join in. Tomorrow's productivity starts with today's experiments.

Takeaways

  1. Start with one bottleneck. Pick a recurring process where time, money, or customer satisfaction leaks away.
  2. Give an agent a task and boundaries. Manage knowledge, access, control, and assign human accountability.
  3. Measure productivity. Look at throughput, quality, cost, and customer impact, not the number of AI tools in use.
  4. Learn yourself. As an entrepreneur, carve out weekly time to experiment with agents and share insights with your team.
← All posts