At a field inspection SaaS, the construction customers wanted the advanced analytics capabilities but, the tooling was sophisticated and brittle. What they actually wanted was the SaaS provider to set it up and maintain it for them. 

The feature was desirable; operating and maintaining it was not.

For a marketing SaaS platform, the enterprise entertainment customers wanted managed service. They don’t want seats or a login. They just want the business results.

An IoT monitoring startup built a real-time dashboard. Customers wanted a defensible number for next year’s budget.

Same pattern. Nobody rejected the product. Everyone rejected operating, learning it, maintaining it.

This is the third shift, after seat-based pricing collapsing under the weight of AI and shrinking headcounts and after professional services firms shift from billable hours to monthly subscriptions.

Even in enterprise contexts, someone on the buyer side owns the vendor relationship. The question is how much of their job description is operating the software?  Are they also a user – or do they just own the contract? In my corporate experience, there was an IT role called “Business Relationship Manager.” Their job was to understand the technology needs of the business and manage the vendor relationship — even if the vendor was internal IT. Mostly they managed the contracts and gave the business someone to yell at when the software didn’t work as expected. 

The BRM was never the user. They were the buffer between the business and the product. Increasingly, this is what enterprise buyer-side contacts want to be: the person who owns the outcome, not the person who logs in.

The preference has always leaned toward having someone else operate it. The economics just didn’t support it — managed service is expensive to deliver, and SaaS margins depend on customers serving themeselves.

AI agents change the economics. Before, one account manager could maybe handle 5 large accounts. The vendor charging a steep premium for the white glove attention. Now, with agents handling the operational work; running reports, configuring campaigns, monitoring systems, flagging anomalies. One account manager can handle 20 customers. Maybe more?

The constraint shifted from “how many clients can one person serve” to “how many agents do we need?” 

Today, if the vendor doesn’t absorb the service layer, someone else will. An agency, a consultant, an intern, a homegrown AI wrapper — or a VC-backed startup purpose-built to sell the outcome directly. Sequoia just published a thesis on exactly this: for every dollar spent on software, six are spent on services. They’re betting the next wave of billion-dollar companies will capture the work budget, not just the tool budget. The question for existing SaaS vendors is whether they’re the ones doing it.

Some vendors have made this work strategically for decades. Microsoft, Oracle, NetSuite, etc. All built enormous partner ecosystems precisely to outsource the managed service layer. The vendor keeps platform revenue; the partner captures implementation and ongoing service margin. It’s a deliberate architecture, not an accident. The difference is they designed for it. SaaS companies losing margin to agencies and consultants today didn’t choose this — it’s happening to them by default.

Product companies don’t strategically decide to become service companies. They drift there, one big logo at a time — tracing one customer’s coastline until the product perfectly fits one and everyone else with increasing awkwardness. Absorbing the service layer with agents is a controlled descent: going deeper into the customer’s shape than competitors, deliberately priced. Agents make it easier to descend, to serve at scale. They don’t make it easier to know when to stop.

The marketing platform I talked with is rebuilding around this exact model. An agentic layer where enterprise clients get managed service experience and smaller clients get self-serve at credit-card-size fees. Not two different products. One product with different fulfilment tiers.

In this Service-as-a-Service world:

• The dashboard is for internal use. 
• The analytics are for internal use. 
• The exception notifications are for internal use. 
• The software is for internal use. 
• The customer just wants the outcome.

Price accordingly.