Services-as-Software: the whole thesis in one page
Every services business is a data business that hasn’t been rewritten yet. Here’s why the economics invert — and how a studio compounds by building one company per industry.
The margin trap of selling hours
A traditional services firm sells time. Revenue scales linearly with headcount, margin stays pinned to labor cost, and the best people spend their days on work a system could do. For a century that was simply the cost of delivering expertise. It no longer is.
Services-as-Software is the inversion: take a workflow currently delivered by people — bookkeeping, company formation, patient intake, product merchandising — and rebuild it so software does the labor and humans supervise the exceptions. The customer still buys an outcome, not a seat license, and the provider’s margin stops tracking headcount.
Why now — and why most attempts stall
The enabling technology is finally good enough, but the failure rate is brutal. MIT’s NANDA initiative studied 300 deployments and found roughly 95% of enterprise GenAI pilots delivered no measurable P&L impact — only about 5% reached production.¹ The gap wasn’t model quality; it was that most tools never fit a real workflow or learned from feedback.
The pilots that crossed over shared a shape: a narrow vertical, deep workflow integration, back-office work where the ROI actually lives, and a provider who owned the outcome rather than shipping a demo.² That is a description of a company, not a feature.
You don’t cross the divide with a better model. You cross it by owning one workflow end to end.
Why a studio, not a startup
If the winning unit is one vertical rebuilt end to end, the winning structure is a studio that can do it repeatedly — shared engineering, brand, GEO and capital, pointed at a new industry each time. Each company compounds the last: the intake agent built for dentistry informs the one for legal; the formation stack behind FormBridge accelerates the next regulated vertical.
That is the whole Gigaverse thesis on one page. One senior team, one operating system, one company per industry that was waiting to become software.