Blog
arrow-right
If you've ever automated anything with AI, you know what the 15% is

If you've ever automated anything with AI, you know what the 15% is

Services vs Software (inspired by services as software companies and AI-native services firms)

A few weeks ago, we took on a healthtech client for a positioning project. After the ICP workshop, the founder dropped a link in our Slack: a YC post about a company called Plena. This, he said, is what I want to become.

Plena sells a solution to medical clinics. To put it simply, it's agentic AI that handles administrative tasks inside the systems clinics already use. 

Their pitch comes down to four words: “We do the work." I always pay attention to messaging. And what I see here is very different from "We are X that allows you to do Y” (the usual software story), where the subject is "you,” "we” are only a tool. But when you say "We do the work,” "we” is the subject. 

Software becomes a service. It's exciting! But there is a small thing that should worry businesses looking to adopt this model.

It's the 15%.

In today's newsletter:

  • What's the 15%?
  • What's next for both service firms and product companies?
  • A word on positioning

What's the 15%?

There's always something that doesn't conform. It can be an odd row in your spreadsheet, a client who runs analytics in a system nobody accounted for, or a workflow that's almost standard except for the one step that isn't. 

15% doesn't fit.

An AI agent handles 85% of the work perfectly well. It's the easiest work that a human can handle just as well, and pretty fast, too. 

But if you're not careful with those 15%, your promise isn't working.

Why "becoming a service" changes your economics

The difference between the old model (buy our tool) and the new one (we do the work) is 15%.

A SaaS company can ignore it because if the software doesn't fit the customer's edge case, that's the customer's problem. They operate the tool they bought and nobody promised them that the tool was going to do the work for them. A services company, on the other hand, can't ignore the 15%. 

So how do you handle it? What's the solution? Hire humans, of course. They can sort out those weird rows and nonstandard clients. This means you need a bigger team, which also means your margins will get thinner, and your economics won't be like SaaS. Because "services as software” isn't a standardized solution.

One SaaS product differs from another through its feature set. "Services as software” differ from each other by operations. Read: the ability to deliver the messy part reliably, again and again, without it falling apart.

Everybody knows that usually, if a services company isn't operationally mature, when it grows, the quality of the provided services drops dramatically. That's normal. Only a few services firms reach operational maturity (and only about 30% survive 10 years).

The same applies to “services as software” companies. If they scale before becoming operationally mature, they won't be able to maintain quality across their client base. The easy 85% scales but the hard part is what cracks under volume. 

So the challenge of this new category is a contradiction. You want to be as scalable as SaaS, while covering the 15% like a services firm. 

The startups that win will look less like high-margin software and more like exceptionally well-run operations.

What this means for both services firms and software startups

Software companies turning into services have to inherit everything services businesses have always known: 

  • Outcome is expensive to deliver
  • Your exception rate is your reputation
  • "We do the work" is only as good as your messiest project 

Services firms turning into software (AI-native agencies and productized consultancies) can finally scale the 85%. But if Claude Code convinced them the 15% disappeared, they've just automated their way into broken promises at volume. (I'm looking at you, content agencies building content generation pipelines to produce crap content at scale).

A word on positioning

The second you say "we do the work," your positioning has to be built on proof of delivery. In other words, an outcome. The thing with outcomes is that they're cheap to promise and expensive to keep.

The agencies that stay order-takers (“tell us what you want and we'll write it”) can never promise a result, but they get blamed for results anyway. 

The ones that own the outcome have to take responsibility for the strategy, and not just the deliverable. 

Software is now walking into the same fork.

So what result do you own? What work do you make disappear? Can you prove you'll deliver it (including the ugly 15%)?

Read my detailed article about this "Software Is Now Plumbing."

See you next week

I feel like the market's changing, the tide is turning for both software and services. It's interesting to watch.

Kateryna

P.S. If we aren't connected already, follow me on LinkedIn and Instagram. If you like this newsletter, please refer your friends.

P.P.S. Need help with quality content? Zmistify your content with Zmist & Copy.

Read also
blog thumbnail
people
I got back a week by automating one of my core workflows

My monthly client reports now write themselves

blog thumbnail
strategy
How to Apply Jobs To Be Done to Content Writing

Here is a question: Does your content help people do a job people are hiring it for? Right, people hire content as well as they hire doctors, plumbers, presidents, or software developers. Let's find out why they do it and how you can use it in content marketing.