Service-as-a-Software
How AI lets service companies deliver faster, better service by freeing their talent from unnecessary work.
The idea
Service companies sell the work of their people: tax advisors, consultants, lawyers, agencies. Yet a large part of that talent's day isn't spent on the service at all. It goes into the tasks around it: collecting and sorting documents, entering data, writing status updates, formatting reports, chasing information, internal admin.
None of this is what the customer pays for. It's time the talent loses.
With AI, a service company can take these tasks off its talent's plate. Software built specifically for their workflows handles the unnecessary work, so the talent can focus on the actual service. The result:
- Faster service, because the work around the service no longer slows it down.
- Better service, because the talent puts their time and attention where their expertise matters.
- Time back for the talent, freed from tasks that have nothing to do with the service they provide.
The question is how to structure a company so this happens systematically, not by accident.
The four layers
The model has four layers. Customers sit outside, and below them is the company as an inverted triangle. The width of the triangle shows how many people work on each layer: the most at the top, the fewest at the bottom. Each lower layer makes the one above it stronger.
Customers
receive the service
They pay for the service. They are connected to the talent directly, or possibly through software in between.
outside the company
Talent
delivers the service, supported by software
The experts who do the actual work. Their work is amplified by software tailored precisely to how they work.
most employees
Developer team
builds the software in-house
It sits close to the talent and builds the tools that make them faster and better. The software is built internally, not bought.
far fewer employees
Developer enablement
enables the developers
A small team that helps developers build software for the talent, with platforms, templates, AI tooling and shared building blocks.
fewest employees
Two models
Where exactly the software sits can vary. In both cases it's built by the developers; the difference is who talks to the customer directly.
Model 1
Software is the interface to the service
e.g. tax returns
- Customer
- ↑↓Software← Developers
- ↑↓Talent
Model 2
Software is a tool for talent
e.g. consulting
- Customer
- ↑↓Talent
- ↑↓Software← Developers
Model 1: Software is the interface to the service
The customer talks to the software, and the software talks to the talent. It's the surface through which the service is ordered and delivered, while the talent works in the background. This fits services that can be standardized well, such as tax returns: the customer uploads their documents, the talent reviews and processes them, and the result comes back through the software.
Model 2: Software is a tool for talent
The customer talks to the talent directly, and the software is their tool in the background. This fits services where the personal relationship and the talent's judgment are the product, such as consulting. The customer may never see the software, but notices that the talent works faster and better.
Why this fits the AI era
AI makes building software cheaper than ever, and it can take over many of the tasks that used to eat up the talent's time. That shifts the bottleneck: writing code is no longer the scarce part, knowing which tasks waste the talent's time and how to remove them is. That knowledge lives with the talent themselves.
A company structured this way uses both. The talent knows where their time gets lost, the developers quickly build the software that takes it over, and enablement makes sure this loop gets faster with every round.
The original notes
This is how the idea started, as sketches in a notebook. Click a page to enlarge it.