# Derek Callahan, Staff Engineer at Patchwork (82 people, Series B) — read of claude-code-assistant, June 20 2026

> 11 years writing backend Python and Go, currently daydreaming about going indie while pretending to be in a sprint planning meeting.

## How I got here

Someone in the Indie Hackers Discord dropped a link to "Wishdeal Factory" and called it "an idea marketplace that actually shows you the math." I'd been poking around Acquire.com and similar sites looking for something half-built I could take over. This showed up in my tab stack and I finally clicked it on my lunch break.

## What I clicked first

The hero pulled me in with "A developer-first AI assistant that understands your codebase" and then I immediately got suspicious. That sentence has been used by Copilot, Codeium, Cody, Cursor, and probably thirty Chrome extensions I've uninstalled. But I scrolled anyway because the framing was different -- this isn't the product page, this is someone selling me the *idea* of building the product. That pivot matters and I didn't catch it until I hit the pricing.

## Where I paused

The Fermi math section stopped me cold. Year-1 take-home: **$-34,000**. Meaningful-success odds: **1 in 9**. I've read a lot of "idea marketplaces" and exactly none of them lead with negative projected income. That's either brutally honest or a very clever trust move. I sat with it for a minute. The financial upside axis is scored 2/10. They published that. That's the part I keep coming back to.

## What I distrusted

Three things.

First, the feature list reads like a Claude API documentation tour, not a product. "Extended Context -- 200K token window captures entire projects." That's just...Claude. That's the model. What's the wrapper doing? "Language Fluent -- From Python to Rust to Typescript." Again, that's the model. None of these bullets describe a product decision, a UX choice, a workflow, or a technical edge. I have no idea what the actual software does or looks like beyond "VS Code extension."

Second, "Stop Waiting for Copilot to Understand" is a tagline that assumes the reader's pain without proving it understands the market. Cursor exists. It's funded, fast, and has 100K users. The competitive landscape here is genuinely brutal and the page doesn't acknowledge it once.

Third, the hero image is described as "Developer at desk, IDE with Claude assistant panel open." Stock image territory. I can't see the actual UI.

## What would convince me

I want one real Loom walkthrough of the VS Code extension doing something Copilot can't. Not a list -- a demo. Specifically, show me the 200K context actually reasoning across a 20-file refactor without losing the thread. That's the claim. Prove it with a screen recording, messy real codebase, no cherry-picked snippet.

Also: the dossier says it includes "first 7 build tasks." I'd want to see the task list before I buy it for $5 to know if it assumes I'm starting from scratch or if there's some scaffolding. The gap between "here's a plan" and "here's working code you can actually run" is where most of these idea packages fall apart.

## What I'd ask in an email reply

1. The feature list is basically "Claude with a VS Code extension." What does the dossier say about why this beats Cursor, which already has deep repo context and is well-funded? Is there a differentiation thesis in the $5 unlock?

2. The year-1 Fermi shows negative take-home. What's the model for how that turns positive -- is it purely volume, or does the dossier describe a specific wedge into a customer segment that's underserved by the current tools?

3. Has anyone who bought the dossier actually shipped something? Even one person. I don't need a success story, I just want to know if anyone's reached "paying customer" on any of the Factory ideas.

## Verdict: on-the-fence

The honesty about the negative projections and 1-in-9 odds is genuinely disarming -- I've never seen a pitch do that and it made me read longer than I would have. But the product itself is almost entirely undescribed beneath a layer of AI-capability claims, and the market it wants to enter already has well-capitalized players. I'd pay $5 to see if the dossier has anything original to say about the wedge. I would not pay $99 or spend 6 months building this without a much clearer answer to "why would someone use this instead of Cursor."

---
*Memo by skeptic persona, generated 2026-06-20. Studio breaks own self-grading loop.*