← essays

dude, where is the roi?

by maurice kleine · · first posted on x

For the past few months, you’ve followed all the steps.

You listened to the right podcasts.
Read the right newsletters.
Followed the right people.

You diligently, nay vigorously, nay obsessively tried every hype.

MCPs. TOON. Ralph. Opus 4.5. GPT 5.2. Skills. Opus 4.6. OpenClaw. Kimi K2.5. Minimax M2.5. GPT 5.4.

You’ve subscribed to the Plus, Pro and Max plans.
You tried the harnesses.
The coding apps.
The skills.
The Claws.
You bought a Mac Mini.
Spun up a VPS on Hetzner.

You made sure to have opinions, to participate in the marketplace of ideas, to reply to the right people, to show everyone how on top of it you are.

You started projects. Abandoned them. Started new ones.

Just to try new tools of course, not because you lacked vision or direction or got frustrated with said tools.

Look at you, you’ve done it all!

And yet,

you’re stuck at $500 MRR and paying the bills with freelance work.

Where is the fucking ROI?

Over the past few weeks, I’ve started to question the ROI of how I’m spending my time, and frankly, money. I tried out every tool, made sure I kept up to date with everything happening on X, of getting the Pro Max Plus Ultra plans.

The truth is, I’ve been really enjoying AI coding. My work for years now has been to ship POCs. First for freelance clients, then more recently for my own ideas.

AI is really good at this! So I started to feel like I had unlocked a superpower.

Ralph loops.
OpenClaw on Discord.
Claude Code Max 90 and then Max 200.
Codex.

Let’s burn tokens baby!!!

But now I’m working on a longer term product and frustration is creeping in.

It’s not just that the code is sometimes outright sloppy, it’s also that I don’t even have the mental capacity to keep track of 10,000 new lines of code being introduced every day. I can't keep track of what changed, and I'm starting to feel increasingly overwhelmed and frustrated.

At the same time, slowing down feels wrong. The promise of “AI coding goes BRR” has been sold to us so many times, and there is constant pressure:

Brain goes:

I, ngl no cap fr fr, also participated in pushing this narrative, and if that caused you stress, I am sorry.

More and more, the promise of being able to build faster than ever before with AI is starting to sound like a ‘get rich quick scheme’. Just throw more money at it, bruv, and AI will make all your builder's dreams come true.

But again, the question pops up.

Where is the mother effin' ROI?

To be clear, AI is great for a lot of reasons. We don’t have to throw out the baby with the bathwater here.

But hype-sensitive minds like mine need to be very aware that a lot of the narrative, especially on X, is not an accurate reflection of the truth.

Yes, AI is being adopted at a crazy scale.
Yes, people are actually losing their jobs because of it.
Yes, the technology is very powerful.

But the "just burn more tokens" strategy is not a business model.

For some people it works.
For most of us, not so much.

Tools do not create outcomes.
Focus does.

More than ever, we need to be aware of the noise by being critical of the (hypothesized) ROI of the latest hype.

So, for the next few weeks, I’m going to attempt to shift my mental model. Less time on X and trying the latest thing, more time on bringing back a more pragmatic, less hype-pilled state of mind.

More focus, more planning things out properly, manually. No PRDs to be fed to an AI model, but rather, writing things down in markdown files and code by hand.

I’m still going to use AI to help me ship, it would be stupid to completely stop using AI.

Tab-completions are great.
Bigger refactors are perfect use cases.
Small features, tweaks, or improvements are fine.
Starting from scratch in a manually scaffolded codebase to quickly go from idea to POC is awesome with the help of AI.

But letting AI rip for hours on a single feature?

Right now, that’s a net negative ROI.

~ Maurice