Vibecode your tools instead of buying them

A small company easily accumulates fifteen software subscriptions. A leave tracker at eleven francs per user per month, a contact form at thirty, a dashboard at eighty. One by one each looks negligible. Added up over three years, you are talking tens of thousands of francs for features you use ten percent of.

That calculation was clear-cut as long as building an equivalent took several weeks. It no longer is. This is probably the most concrete change vibecoding brings to small organisations — more so than faster product development.

What the maths became

The classic reasoning set a predictable purchase cost against a high and uncertain build cost. The second rarely won. Today a narrow-scope internal tool — a form, a tracker, a dashboard fed by an existing database — takes a few hours.

The shift is not in the cost of building but in the break-even point. A tool that needed fifty users to justify itself can now justify itself for three. That is what makes it sensible to build things nobody would have considered building.

What replaces well

The best candidates share three traits: a narrow scope, data you already own, and no particular regulatory obligation.

• Forms and their tracking — leave requests, expense claims, room bookings. Stable scope, simple rules, no need to evolve.
• Internal dashboards aggregating data you already hold. Most of the price of a visualisation tool buys flexibility you have no use for.
• Automation of repetitive tasks — rename, convert, notify, chase. This is the historic use case for a script, now available without knowing how to program.
• Domain tools with no market equivalent, the ones you currently run in a shared spreadsheet for lack of anything suitable.

That last point deserves attention: those pay off most. You are not replacing a subscription, you are removing a spreadsheet six people email around in three diverging versions.

What you should still buy

The symmetry is sharp: as soon as a tool carries legal responsibility, an availability obligation, or complexity that evolves without you, the subscription is the right call again.

• Payroll and accounting. Rules change every year and mistakes cost fines. You are not buying software, you are buying somebody tracking legislation on your behalf.
• Authentication and identity management. Rolling your own password handling is the textbook false economy.
• Payments. Card industry compliance is not something you vibecode.
• Anything that must work while you sleep. An internal tool that goes down gets fixed tomorrow; a customer-facing service does not.
• Anything handling health data or sensitive personal data, where the bar sits far above the cost of the subscription.

The cost the calculation forgets

Comparing two hours of generation against a monthly subscription is dishonest. An internal tool carries a cost of ownership the subscription includes and everyone forgets.

• Hosting, modest but real, and certificate renewal.
• Security updates for dependencies, which do not wait for your availability.
• Backups, and above all restoring them — a backup never tested is not a backup.
• The human factor: if one person alone knows how the tool works and they leave, the tool leaves with them.

That last point is the most underestimated in small organisations. It is not solved by exhaustive documentation, but by a short note explaining what the tool is for, where it runs and how to restart it.

Deciding in four questions

1. What does the subscription cost over four years, all users included? That is the only figure worth comparing.
2. What happens if the tool is unavailable for a day? If the answer is "not much", building it internally is on the table.
3. Do the rules it applies change without me? If so, buy: you would be paying for that monitoring out of your own time.
4. Could anyone other than its author take it over? If not, this is not a saving, it is a deferred cost.

One last thing: start with a tool whose failure costs nothing. You will learn the real cost of ownership on a case where you can afford to discover you underestimated it.