If Building Software Gets 10× Easier, How Do You Keep It From Getting 10× Less Controlled?

There's a real shift underway for any organisation that has spent years constrained by software capacity. The bottleneck is loosening. Where you once needed a large team or an expensive supplier engagement to build anything substantial, AI now lets a small internal team build far more, far faster. For a business that has always had more software it needed than capacity to build it, this is real, and it's transformative.

It also contains a trap, and the trap is the mirror image of the opportunity. The same force that lets you build ten times more software lets you build ten times more software that nobody agreed to, nobody documented, and nobody can govern. Capacity and control are not the same thing, and AI increases the first while doing nothing for the second.

The capacity you gain and the control you can lose

Think about what happens when building gets easy. More people across the organisation can commission software. More gets built, faster, in more places. Individual teams solve their own problems without waiting for a central queue. All of that is upside. But every one of those newly-easy builds carries the same questions it always did:

• Why was this built?
• What was it supposed to do?
• Who approved it?
• What constraints apply?
• Does it meet them?

When building was slow and centralised, those questions got answered almost by default, because the friction of building forced a degree of deliberation and oversight.

Remove the friction and you remove the accidental oversight with it. Software starts appearing faster than the organisation's ability to understand what it's accumulating. You get capacity you can measure and a control gap you can't. And that gap is where the real cost hides: not in the building, which got cheap, but in the growing sprawl of systems that exist without agreement, without recorded intent, without anyone accountable: a portfolio the organisation owns but doesn't know.

Control is not a brake on capacity. It's what makes capacity safe to use.

The instinct, when people hear "control," is to imagine a slowdown: approval queues, governance boards, the very friction AI just removed. That's the wrong picture, and it leads organisations to treat capacity and control as a trade-off where they have to pick one. They don't. The organisations that will benefit from AI-scale building are the ones that keep control without reintroducing the friction — where the record of what was agreed, why, and by whom is captured as a natural part of building, not bolted on as a gate afterwards.

That means the goal isn't to build more software. It's to build more software while retaining the knowledge and governance that lets you keep understanding, changing, and standing behind what you built. Capacity without that is not an advantage. It's a liability accruing quietly, one ungoverned system at a time, until the organisation can no longer answer basic questions about its own software estate.

The 10× enterprise, not the 10× developer

Most of the conversation about AI and software is about the individual: the ten-times-more-productive developer, the engineer who ships in a day what used to take a month. That's real, but it's the small version of the story. The large version is organisational. The opportunity isn't a faster developer — it's an enterprise that can build far more of its own software, depend less on external suppliers, and move faster as a whole, provided it can do so without losing its grip on what it's building.

That proviso is the entire game. A ten-times more productive developer inside an organisation that can't track what's being built just produces the sprawl faster. The ten-times enterprise is the one that couples the new capacity with retained control — that can build more internally, direct implementation to whichever mix of internal teams, suppliers, and AI agents makes sense, and still hold one clear, owned record of what was decided and whether the result matches. Capacity gives you the speed. Control is what lets you keep the gains instead of drowning in them.

The question every organisation should be asking now

If your organisation is starting to feel the new capacity — building more, faster, with smaller teams — the right question is not "how do we build even more?" It's the harder one: as building gets easier, how do we make sure it doesn't get less controlled at the same rate? Because the two move together by default, in opposite directions, and closing that gap is not automatic. It's a choice, and it's the choice that separates the organisations that will compound their AI advantage from the ones that will spend the next few years untangling what they built when it got easy.

Build more. Just make sure the organisation still knows what it built, why, and whether it can stand behind it. Everything else AI gives you, you can lose again in the systems nobody can account for.