Software vs. Services

I mentioned a few weeks ago about how I don’t see myself as a SaaS customer in most cases these days.

But that doesn’t mean software is dead at all, just that it needs a facelift and to start selling work again.

Let me explain…

But first I’m kicking off a new weekly section of the newsletter.

Each week I’ll share a new skill I built and use internally at Rye Valley.

Totally free of course.

After last week’s AI MBA skill got hundreds of downloads I figured let’s keep it rolling.

This week’s skill is called Sell The Work. It helps your AI figure out what outcomes you could sell and how to price them.

• You tell it what you (or your software) actually do today

• It helps you pick the finished job a customer would actually pay for

• Then it works out what that job costs you to deliver and what to charge

If you try it, reply and tell me what work you’re going to sell.

The Promise of SaaS

Software has always replaced human labor.

Before the dotcom era, massive IT teams managed on-prem servers.

Then SaaS (when Salesforce launched in 1999) cut that down to a login. The vendor ran the servers for you.

This was the pitch.

You had the tool (software) and your human(s) could use it to build with.

Think of it like giving them a hammer.

In this model, it made sense to charge based on how many people you were giving hammers to.

The Hammer vs the House

LLMs changed the game.

All of a sudden you weren’t being provided a hammer anymore.

Instead you got a house (the exact thing you used to use the hammer to build).

So everyone’s perspective has since shifted, and now they’re thinking why would I buy a hammer when I could buy an architect, general contractor, and labor crew to build a house for me at a comparable price?

It’s obvious why traditional SaaS has been devalued.

Traditional SaaS took no responsibility for the quality of the work. It just provided the tool and let the human operator be the execution and accountability layer.

But now, when you hear people like Sarah Tavel talk about “selling the work,” what they’re really saying is that software is expected to do the work itself, and to own the quality of it too.

And in this new world, charging per hammer gets weird. If your software is doing the work, your customer should need fewer people over time... which means the better your product gets, the fewer seats you sell.

Software Can Sell the Work Too

Predictably, I see a lot of people saying software is dead.

It’s not, but we do need to rethink what and how we’re selling.

Intercom is already selling the work. Its AI support agent, Fin, charges $0.99 per resolved conversation. If Fin doesn’t solve the problem, you don’t pay.

So maybe LLMs are just a better hammer, and one person can now build a whole city instead of a house.

Or maybe the software should just build the house.

Usage vs Outcomes

There are two paths forward for founders building software now…

The Better Hammer (Usage-based Pricing)

If you believe LLMs are just a better tool, you sell productivity.

Charge based on data processed, API calls, or tokens used.

This is software that helps one human build a city.

Your risk, though, is that you only get paid when a human picks up the hammer. And as models keep getting cheaper, it’s harder to charge much for usage alone.

The General Contractor (Outcome-based Pricing)

If you believe software should sell the work, you stop charging for the tool and charge for the result instead:

• A CRM becomes booked meetings

• Accounting software becomes closed books every month

• A help desk becomes resolved tickets

Personally, I think the second one is the better bet right now.

When software takes responsibility for the outcome, you get to charge against the payroll budget instead of the software budget. And, for most companies, payroll has historically been way bigger.