# Marcus Delgado, Staff Engineer at Fieldstone Labs — read of Code Morph, 2026-05-23

> 11 years writing backend TypeScript, been burning nights on side projects for 3 of them. Two attempts. One fizzled at $0, one makes $180/mo. Currently in that restless phase where I'm scanning for the next one.

## How I got here

Someone in the Indie Hackers Discord dropped a link and said something like "this scoring transparency is actually interesting." That's enough bait for me. I clicked expecting yet another launch platform with fake testimonials. I was not fully right and not fully wrong.

## What I clicked first

"Your pseudocode, in your code style" hit me immediately because I've actually felt this pain. The specific thing that landed was "RAG-Aware Learning — Each translation you accept feeds back as context. Your codebase becomes the prompt." That's not just a vibe. That's a mechanism. I know how RAG works. That sentence actually says something. Most hero copy on tools like this says "blazing fast" and "context-aware" and I glaze over. This one was specific enough that I kept reading.

## Where I paused

The honest disclosure block. Specifically: "we don't have live customers on this idea yet. We shipped the strategy package; you ship the customer conversations." I read that twice. So this page is selling me the right to build this product, not the product itself. The hero copy behaves like I'm buying a working tool. Then mid-page it quietly tells me I'm buying a dossier. That whiplash is worth sitting with. It's not dishonest, but it's a strange UX for a product that scores its own "landing page quality: 7/10."

## What I distrusted

A few things. One: "pain intensity: 10/10, uniqueness: 5/10" in the same breath. If the pain is a 10 and the uniqueness is a 5, that usually means GitHub Copilot, Cursor, Codeium, and Supermaven have been eating this lunch for two years. The page never addresses that directly. If there's a real wedge against those, I didn't see it argued. Two: the Fermi math is "Year-1 take-home: -$20,990" and "1 in 9 meaningful-success odds." I respect that they printed it. I also notice they printed it without explaining how the model works, which means I can't dispute it or trust it. Three: the pricing page says "$99-$199 to adopt the build" but then lists "working code starter, brand assets, copy library" as what you get. Is the code starter the Code Morph tool itself? A scaffold? A template? No idea.

## What would convince me

I don't need a case study. There are no live customers and they said so. What I'd want is one real Loom or screen recording of the actual tool running against a real codebase. Not a polished demo. A messy one, with a medium-complexity TypeScript file, where I can see what the pseudocode input looks like, what the output looks like after one feedback loop, and what it looks like after ten. The whole "your codebase becomes the prompt" claim lives or dies on whether the output actually starts looking different after training. Show me that drift. That would get me to the $5 unlock and probably to $99.

## What I'd ask in an email reply

1. GitHub Copilot with a custom `.cursorrules` or a repo-level system prompt already does some of this. What does Code Morph do that a well-configured Cursor project doesn't?
2. When you say "working code starter" in the $99 tier, what's the actual state of it? Is the RAG feedback loop implemented or is that the part I'm supposed to build?
3. You scored uniqueness at 5/10. That's a real concern for me. Who do you see as the direct competitive moat, and what's your answer to it?

## Verdict: on-the-fence

The mechanism is described specifically enough that I'd click the $5 unlock if they answered question 1 convincingly. But the page tries to be a product demo and an idea marketplace at the same time, and the seam between those two modes is visible and a little uncomfortable.

---
*Memo by skeptic persona, generated 2026-05-23. Studio breaks own self-grading loop.*