The Services Play

For years the advice was the same: get out of services, build software. Recurring revenue, fat multiples, that's where the money is.

That just flipped.

SaaS valuations are compressing while AI-native service firms command 30x profit multiples. $2 million in profit, $60 million valuation. And Andreessen Horowitz, Sequoia, and Y Combinator are all moving toward services-as-software, the layer between pure software and pure labor where someone runs the thing AND gets paid for the result.

Here's what's going on and what to do about it.

You're leaving money on the table

Here's the economics, and it's worth slowing down on.

For every $1 a company spends on software, it spends roughly $6 more making that software actually work. Implementation. Strategy. The person who configures it, runs it, and fixes it when the campaign breaks. The tool is $1 of the budget. Everything around the tool is $6.

SaaS companies can only ever capture the $1. That's the structural ceiling of the model, and it's the bucket every software vendor on earth is fighting over. The $6 sits right next to it, mostly uncontested, because software companies don't want to deliver services and traditional service firms can't deliver them at software margins.

That's the gap AI just closed. A services firm running agentic delivery can now capture the $1 and the $6. The whole $7. You're not selling a tool and hoping they use it. You own the software layer and the outcome it exists to produce.

Six times the market, and the incumbents on both sides are structurally unable to chase it. That's why the multiples are moving.

Stop selling hours. Sell loops.

Most service businesses sell hours, then deliverables, then expertise. Each one's a step up, but they share the same flaw: the unit of value is human effort, and human effort doesn't scale predictably. One client gets your A player. Another gets your C player. Same documented process, wildly different results. This is why people hate agencies.

A loop fixes that. It's an end-to-end business process that turns inputs into outcomes over and over, and gets smarter from feedback. Run the ad variants, watch the data come in, scale the winners, kill the losers, save the learnings. Then it runs again, except now it starts from everything it learned last time.

An agent solves one problem. A loop compounds. You need both, but the loop is where the value lives.

Your org chart has to change

The old org chart is built around functions. Account manager, strategist, SEO specialist, designer. People reporting to job titles instead of outcomes.

Flip it.

You need outcome owners who treat their revenue channel like it's their own business. Loop owners who run the loops day in, day out. Agent designers who keep tweaking the specialist agents inside those loops. And one executive sponsor who makes the hard trade-off calls when priorities collide.

Humans don't disappear in this model. They just move up the stack. You hire them for judgment, strategy, and taste, and you let the agents handle the robot work.

What changes your offer stack

Menu of services is dead. It always was a weak position to be in.

The new offer stack:

• Revenue channel management. You own a channel against agreed indicators. Outcomes, not hours.

• Marketing infrastructure buildout. Security, provisioning, governance. The unsexy stuff that separates partners from vendors.

• AI-native marketing transformation. Executives need growth operation systems, not a bunch of new tools nobody uses.

How to start

Pick one loop. Just one. The highest-impact, simplest-to-execute one you've got. Build it with a specialist agent inside, put a human on it, and let it run for two weeks. Look at the data. Improve it. Run it again.

Don't try to transform the whole business at once. Make one loop compound, then move to the next.

The $1 trillion opportunity sits in the layer that makes products actually work for real businesses. That's the whole game.

If you want to see the full breakdown including the org chart, watch it here: https://www.youtube.com/watch?v=zgoztAZuJP4

To building better loops,

Eric Siu