For years, mid-sized companies kept two lists. One was everything their software should do. The other was what they could actually afford to build. The gap between them is where good ideas went to die. AI just closed most of it.
A client recently asked us for something small but awkward: customer-specific price lists they could download as Excel or PDF, and a way to turn any price list straight into a filled cart. The old way would have been a few days of work and two developers with different specialisms. We built it in four hours. Faster and better. This client never had the budget for the old version, so in the old world it simply wouldn’t have been built.
That’s the real shift. Not that custom software got a little cheaper, but that an entire category of “nice, but too expensive” is now firmly in the “worth building” pile.
A bigger example. We recently built a product-configuration platform for Odoo eCommerce: the kind of tool that lets a customer click together a made-to-order product and see the right price on the spot. Half a year earlier, another company had bought a comparable system from a third party: around €2,000 a month, plus a setup bill in the tens of thousands. We deliver the same thing for less than a tenth of that price. It runs at cpqbuilder.com.
Same pattern, bigger number. Software that used to be the preserve of companies with deep pockets is now within reach of a forty-person business.
The same thing happened with the cloud
We’ve been here before.
One morning in October 2007, the neighbours above our Amsterdam office all decided to shower at once during a renovation. The ceiling came down. Water everywhere. While we scrambled for an expensive backup location, I made a decision that reshaped my career: I left traditional IT behind and went all in on something most of my peers still distrusted, the cloud.
A few years later I wrote a book about it, Cloudsource je ICT!. The message was simple. Stop owning servers, rent computing as a service, spend the savings on your actual business. The technology was ready. The objections were not.
And they were loud. Security, mostly. At the time, 47% of IT managers saw the cloud as a security risk, and fewer than a third thought the benefits outweighed the dangers. A lot of people just couldn’t get comfortable with the idea of their data living “somewhere else.” I used to point out that a server in your own basement gives you the feeling of safety, not actual safety - the same feeling people had about internet banking right before they all started doing it.
The part worth remembering: most of that fear didn’t come from the businesses themselves. It came from the IT people whose income depended on the old model. The more servers a company owned, the more work there was for them.
The objections to AI sound exactly like the objections to cloud
I’m hearing the same objections again now, in 2026:
- “The cloud isn’t secure” has become “AI-written code isn’t safe.”
- “I don’t want my data where I can’t see it” has become “I don’t understand the code, so who’s in control?”
- And the objection no one says out loud: the people keenest to tell you AI-built software is dangerous are often the ones who bill by the hour to write code by hand.
The worries aren’t imaginary. But we’ve run this experiment before, and we know how it ends: the technology wins, most of the early fear turns out to be fear of the unknown, and the few real risks get handled by people who take them seriously.
When custom software isn’t the right answer
Nobody who sells you something will say this, so let me: AI-built custom software isn’t always the answer.
If standard Odoo already does the job, don’t customise it. Every line you add is a line someone has to maintain. If your process is a mess, software won’t fix it - fix the process first. And when a requirement is genuinely core to your business, that’s where you want experienced people reviewing every decision, not just a quick AI prompt.
The shift isn’t “build everything now.” It’s that the bar for “worth building” dropped. A lot of things that sat just below it now sit comfortably above.
Who makes sure it works, and stays secure?
Fair question. If AI writes the code, who makes sure it actually works, and who keeps your data secure? That deserves a real answer, not a reassuring wave of the hand. It’s also the question I get asked most. The short version: AI is the engine, but we’re still the ones who build it, test it, and stand behind it. More on exactly how we keep AI-built code clean and secure soon.
If you run a mid-sized company and you have a shelved wishlist - the integration you gave up on, the portal you couldn’t justify, the report that still lives in someone’s spreadsheet - it’s worth taking that list back out. The maths has changed. A free Odoo scan is a good place to find out what is suddenly within reach.