# Marcus Reyes, Senior Software Engineer at Fieldstone Analytics — read of CodeMorph, 2026-05-18

> 8 years writing TypeScript, currently the third engineer at a Series B SaaS that just hit 22 people. I own the frontend architecture and half the API layer. I pay for Cursor personally because the company won't approve it.

## How I got here

I Googled "copilot style consistency codebase" after a code review where I spent 20 minutes on naming convention comments that Copilot had introduced. Found a Reddit thread where someone complained about "output drift" and a reply linked here. Not an ad. Just a dude venting.

## What I clicked first

"Write code in your style, not the LLM's" landed. That is the exact frustration I brought to this page. I clicked the live demo area before I read anything else. I was hoping to paste something and see it run.

## Where I paused

The RAG flywheel claim stopped me. "Every accepted translation becomes RAG context. Your second translation is more on-brand than your first. By translation 50, the LLM writes in your exact voice and patterns." That is either the actual insight of this product or it is a sentence that sounds good and does nothing. I sat on it for maybe two minutes. If it works, it is the real differentiator. Every other codegen tool I have tried starts from zero every time. That gets old fast.

## What I distrusted

Okay so then I scrolled and hit this: "Honest disclosure: we don't have live customers on this idea yet." And I had to stop and re-read the whole page.

This is not a product. This is an idea. The hero section presents it like a working tool. The "Try it Live" framing and the before/after block look like a real SaaS demo. It is only buried in the scoring section that I realize I am looking at a strategy package for a business someone could go build. The $99 tier is not a subscription. It is "buy the starter code and go build it yourself."

That framing mismatch is a real problem. I felt tricked, even though the disclosure is technically right there.

Also: "$-20,000 Year-1 take-home (Fermi)" -- negative twenty thousand dollars. I read that three times. I think it means the builder loses money in year one after investment costs. The math may be honest but presenting it in that format, with a dollar sign and a negative, on a page trying to get me to spend $99, is a confusing choice.

## What would convince me

If this were a real tool I could actually sign up for, a 10-minute session log showing real before/after translations on a public open-source repo would do it. Not a polished marketing video. A Loom of someone pasting messy pseudocode and watching the output come back with their actual project's naming conventions intact.

If it is an idea kit I am buying the rights to build: I would want to see one person who bought the $99 package and shipped something, even a small waitlist, with a quote that is not "this gave me clarity." I want "I launched, got 12 paying users, here is what I changed from the plan."

## What I'd ask in an email reply

1. The live demo on the page -- is that an actual running instance I can test, or is it a static mockup? I clicked around and could not tell.

2. The RAG context accumulation: is that stored per-user in your infrastructure, or does the buyer build and host that themselves when they adopt the code package?

3. The "$99 adopt the build" tier includes "working code starter" -- what stack is that, and does it include the RAG pipeline or just the translation layer?

## Verdict: on-the-fence

The core insight around style drift is real and I have felt it weekly. But I spent four minutes on this page before I understood that it is not a product I can use, it is a blueprint I would pay to go build myself. That gap between what the hero promises and what the page actually sells is the thing standing between me and clicking anything.

---
*Memo by skeptic persona, generated 2026-05-18. Studio breaks own self-grading loop.*