Service-as-a-Software
The one thing you need to know in AI today | AI-Ready CMO
Y Combinator just dropped its summer 2026 Request for Startups. We covered the spring edition in February — AI-native agencies operating at software margins. Every CMO has had the agency-margin conversation by now. If you haven’t, well…
The summer RFS goes one layer deeper and deserves a new acronym because the old one no longer means what it used to.
For 20 years, SaaS meant Software-as-a-Service. You paid for a subscription. You got access to a tool. Your team did the work inside that tool.
YC just funded the inversion: Service-as-a-Software. You pay a subscription. You get the outcome. Nobody on your team does the work — because the service itself is now the product. Plug-and-play workforce is the endgame.
The summer RFS describes the three layers that make this possible:
Layer 1 — Company Brain. Extract tribal knowledge from people’s heads, Slack threads, tickets, and old emails. Structure it. Keep it current. Make it executable. Any startup that builds solutions for a company brain will have a very fast exit to one of the enterprise giants.
Layer 2 — The AI Operating System. Wire that brain into the live systems. Every meeting recorded, every ticket tracked, every decision logged. The company stops running on memory and starts running as a closed loop. Say hello to automating half of what any HR, procurement, legal, or finance department does today.
Layer 3 — AI-Native Service Companies. Plug in an external provider that runs the boring 80% of work — insurance brokerage, accounting, compliance, healthcare admin, and yes, marketing operations. It’s not external, like a freelancer, a consultant, or an agency. It’s basically a workforce installed for and in your company. Plug-and-play resource.
These layers are built on top of each other, and they won’t work without perfecting the lower layers first. Read it as three separate categories, and it’s a research note. Read it as one stack, and it’s how Service-as-a-Software gets built.
In Partnership with Am I on AI
You don’t need another dashboard. You need to know what to do Monday morning.
Most visibility tools tell you what happened. That’s useful for reports. It’s not useful for decisions.
What actually moves the needle: knowing which specific activities are driving your AI visibility this week, and which ones are quietly wasting budget.
That’s what the audit gives you. Not a score to screenshot. A direction to move in.
The new contract
The old SaaS contract was simple. You wanted a new capability, say graphic design, drag-and-drop website building, data analytics, and so on, so you signed up for a tool that gave your people that capability. The entire workforce’s career depended on these tools: do you know this, can you work with this, how proficient are you with this?
The old contract was: we sell you a tool, you figure out how to use it. That’s why every SaaS purchase came with implementation costs, training budgets, and a CSM emailing you about adoption.
The new contract has nothing to do with capabilities. It’s only about outcomes. The moment your company has a brain and your workflows are scanned and already partially automated internally, you lose something that made you unique. That pesky and ghastly institutional knowledge, which made everyone feel that their company and work are so unique. It’s not. Almost none of it is unique.
The new contract: we sell you a service that is fully adapted to your company and workflows, and it is mostly run on software (AI). We will send the people who will install this new service for you. All you need to do is pay the subscription and provide the access points for the install.
The reason this couldn’t happen before isn’t that the models weren’t good enough. They were. The reason was that no outside provider could understand your business well enough to deliver real work without your team holding their hand.
Layers 1 and 2 fix that. Once your context is extracted and live, an outside provider can execute against it cleanly. The handholding goes away. The service becomes the software.
The new marketing team
For marketing, the order of operations is:
Someone maps how marketing decisions actually get made at your company.
That map gets wired into your tools and stays up to date.
An external provider plugs in and runs the production layer — campaign briefs, copy variants, asset production, performance reporting — at a fraction of current cost.
Your agency stops being a strategic partner. Your martech vendor stops being a tool. Both become services you subscribe to and outputs you consume. The line between “agency” and “software” disappears, because the agency is the software.
This is also why this week’s Adobe story matters. When Adobe became a backend to Claude, it stopped being software in the old sense. It started being a backend service called via API. Same pattern, different layer of the stack. Service-as-a-Software is the direction of travel for everything in your category.
And on that unique workflow of yours? Super cute, but it is going to become training data. Literally every company is building that, right now. We wrote about what Meta is doing to their own employees. It’s just the most visible example among thousands of others.
The company brain solution with an operational layer is coming. Probably at the end of this year, early next year. Once those two layers are done, the rest is pretty simple and market-driven, so it will happen almost overnight.
Most marketing leaders have their wake-up call when their C-Suite tells them it is time to let go of 80% of their team, because there is this new company that does everything the marketing team does for 1/100 of the payroll costs in expense-ready subscription tokens.
Don’t be that marketing leader.
— Torsten and Peter
Share with a colleague. Earn paid membership months.
Your AI Marketing Tools