In defense of software businesses - by Mert Deveci

SubscribeSign in

In defense of software businesses<br>Software is dead. Long live software.

Mert Deveci<br>May 16, 2026

Share

Are you working hard enough? Did you raise a $650 MM seed round yet?<br>Why? Why not?<br>Maybe you are working on the wrong thing. Maybe you should pivot. Though if you stick 4 years, you might be acquired for $20 Bn. Or go public - then get valued at 1.5x your cash in the bank account.<br>You should work on hardware. Or crypto. It is unsexy now. Is it? How do you even work on hardware? But do it in the ocean, because robotics on ground is easy now.<br>Maybe just build a services company. It is actually a software company - but not really. Replace lawyers - with other lawyers. And sprinkle some AI on top.<br>Why?<br>Well - didn’t you hear? Software is dead now.<br>You should work on literally anything else.

Schrodinger’s market: market where you are not sure if it is the bottom or the top - because it is definitely not normal. At least, it doesn’t feel normal.

Should you just go work at Anthropic? They added $14 Billion since March - and you are pivoting? Can you even work at Anthropic? Isn’t Mythos better than you already?

What to work on if not software?

You can:<br>Build a services company. A law firm. Runs on AI.

Work on hardware. Somehow.

Software

1. Services

So VCs will tell you you should just become the law firm man. “Cut out the middle men, get paid more.” Build one yourself or roll up some existing firms.<br>It is the future of software.

Software is too competitive. There are more than 1,000 AI legal startups out there.<br>You know what is more competitive? Law firms. There are more than 400,000 law firms in the US.<br>But we use AI.

Well - does the customer give a damn? You were supposed to deliver an NDA. Is it cheaper? Yes. Is it faster? Yes.<br>You see - I don’t want it cheaper. I don’t want it faster.<br>I wanna know you put the right amount of effort into it. Because it is really freaking important that we get it right.<br>2. Hardware

I am not the right person to comment on hardware. All I know is that it is really damn hard.<br>However there is one thing I know: Nothing has changed here just because of AI.<br>Hardware is hard man.<br>3. Software

There are two matters to look into here:<br>Who does the work?

Who makes the money?

Who does the work?<br>Anthropic employs around 1,500 engineers today and hiring 450 more.<br>Why? Why hire more engineers if Claude Mythos can literally break the world?<br>You see - because it is software we are all working on. Software is what makes money.<br>And AI is software.<br>LLMs write the code. However they only do so if you send the prompt. Someone needs to still write that prompt, configure the hooks with the right prompt - queue the messages<br>And if you are a good engineer: actually read the code.<br>Who makes the money?<br>VCs are pushing the narrative of becoming a services company for new startups. A new insurance broker, a new rollup, a new law firm<br>Well this is nothing new:<br>A new insurance broker: Marshmallow founded 2017

A new rollup: Thrasia in ecommerce, founded in 2018

A new law firm: Atrium, founded 2017

#2 and #3 are gone.<br>Irregardless, AI-native services are attracting investment now. Take Crosby. AI-native law firm, just raised $60 MM Series B.<br>Needless to say, they are doing well.<br>You know who is doing better: Harvey.<br>A pure, software based, LLM wrapper, enterprise software company. Simply because they are making more money. And they do not seem to be decelerating.<br>I know I know, their software is not that good, it is a sales based organisation, they are virtue signaling and so on.<br>This might lead you to believe that building an AI native services firm is the better way to penetrate this market.<br>Well hate to break it to you - but there is no market on earth that is more superficial in its procurement than services.<br>Let me tell you how in one sentence as someone who worked at Morgan Stanley and Societe General before:<br>Morgan Stanley is one of the Bulge Bracket banks (Goldman, JP, MS)

When I was at SocGen, we would work so hard on every single pitch to clients, drop our prices, deliver the work faster and better

And MS would still get the deal.

This is because professional services, from a sales perspective, sell perception. Not tech, not automation. Perception.

All of this boils down to two points:<br>It is all software that we are working on. Even though it is dead.

In a market where software is unfavorable in perception - it just might be the most contrarian thing to work on for the first time in history.

Software is dead. Long live software.

Share

Previous

Discussion about this post<br>CommentsRestacks

TopLatestDiscussions

No posts

Ready for more?

Subscribe

© 2026 Troy Guild Ltd · Privacy ∙ Terms ∙ Collection notice<br>Start your SubstackGet the app<br>Substack is the home for great culture

This site requires JavaScript to run correctly. Please turn on JavaScript or unblock scripts