Businesses shouldn't bend
to software.

I. The misfit is everywhere, not just in the big systems

When people hear "software that doesn't fit," they picture the seven-figure ERP that never matched the operation. That happens. But the misfit is far more ordinary than that and far more expensive in aggregate.

It's the sales team that redesigned its pipeline to match a CRM's object model instead of the other way around. It's the marketing team running a single campaign across five subscriptions, three automation glues and a shared spreadsheet that reconciles what the tools can't. It's the operations manager who is, in practice, a human API between systems that were never meant to work together.

Each of these looks like a workflow quirk. Together they are a structural condition: process bending to software. And the cost of that condition has a name.

II. Why near-fit became the default

Nobody chooses misfit. They choose the rational option of their era. For twenty years, off-the-shelf SaaS was that rational option: it solved the common 80% of a problem at a fraction of the cost of building and the remaining 20% was absorbed by the organization, quietly, in workarounds.

The vendors built an economy on that arrangement. Per-seat pricing that scales with your success. Customisation quotes that approach the cost of building from scratch. Data held in formats designed to make leaving harder than staying. None of this is a scandal; it's the business model working as designed. But it means the 20% you absorbed never shrinks. It grows with you and the exit gets more expensive every year.

Renting almost-right software is the only purchase a company makes where the product's misfit becomes the customer's job to fix.

III. What changed: the economics of fit

Until recently, the alternative, software built to fit, was a rich company's privilege. Bespoke meant years of delivery and enterprise IT budgets. For everyone else, near-fit was simply the only seat at the table.

AI-augmented delivery broke that math. Small senior teams, orchestrating AI agents for the volume work while keeping human judgment on the decisions, now ship in months what traditionally took years. The build that was out of reach on a mid-market budget in 2022 is, in 2026, routinely cheaper than five years of renting a near-fit, before counting the tax itself.

This is the part most companies haven't priced in yet: the build-vs-buy equation they last calculated years ago has flipped and the spreadsheet they used is stale.

IV. Fit at every scale: Process, System, Platform

The second thing that changed is scale. Fit-built software no longer starts at the platform level. It starts wherever the misfit hurts most:

Process: one workflow, one team, weeks. The campaign ops tool that replaces a five-tool stitch. The pipeline shaped to how your sales team actually sells.
System: an operational system that runs a department, months. The ATS that processes thousands of CVs a day against your ERP. Order management shaped to your fulfilment.
Platform: the thing the business runs on, quarters. The property platform built around your portfolio, not a vendor's template.

Size the engagement by the misfit, not by your company. Most companies should start small: one process, one team, one owned tool, then let the result argue for the next one.

V. What ownership actually means

"You own it" is easy to say. Here is what we mean, concretely: the code lives in your repositories. The infrastructure runs in your cloud accounts. The documentation, architecture records and test suites are handed over with it. There is no licence fee, no per-user pricing and no dependency on us to keep it running. If we disappeared the day after handover, your software wouldn't notice.

That last clause is the discipline. A build partner who keeps a hook in your system (a licence, a proprietary layer, a hosting dependency) has rebuilt the landlord relationship with extra steps. Ownership is only real when the handover is total.

VI. The decision in front of you

None of this means everything should be built. Plenty of near-fit software is good enough and the honest answer is often "stay." The point is that the question deserves real numbers: what the current misfit costs you per year, what closing it would cost once and which way the five-year math points. That's a two-week diagnosis, not a leap of faith.

The question is no longer "can we afford to build?" It's "can we afford not to own?" And for the first time, it has a defensible, calculable answer.