# Derek Fonseca, Independent Software Founder — read of AI Agent Cost Monitor, June 13 2026

> "14 years writing software, 2 small SaaS products shipped, currently running a Postgres-backed B2B tool for property managers. Third attempt at finding a side product I can actually sell."

## How I got here

I was scrolling through the Indie Hackers forum on my phone at my kid's soccer practice and someone posted a thread titled "anyone tried buying a Wishdeal idea yet?" A few comments said the honest scoring was interesting. I searched for the site on my laptop that evening, landed on their ideas index, and clicked this one because I had literally just gotten an OpenAI bill that was $400 higher than I expected last month from a batch job that looped badly.

## What I clicked first

"Stop wasting thousands on runaway AI agent API calls" got me. That's the right sentence. I didn't skim past it. The problem is real. I just lived a version of it. So I kept reading.

## Where I paused

The Fermi math: "$-22,500 Year-1 take-home." Negative. Not zero. Not optimistic and wrong. Negative. I've seen a hundred of these "here's a startup idea" pages and I have never seen one lead with a projected loss. I sat with that for a minute. Either this is the most honest product studio on the internet or there's something buried in the math that makes this look worse than it is. I genuinely wasn't sure which.

## What I distrusted

This whole page describes a product that does not exist. The feature list reads like a working app: "Live Dashboard," "Auto Kill-Switches," "Weekly Cost Report." But then I hit: "Honest disclosure: we don't have live customers on this idea yet. We shipped the strategy package; you ship the customer conversations." So the features aren't a product. They're a scope document. That's fine, but the layout made me read the feature table as if I was evaluating software I could use. I had to reorient mid-page. That friction is on them.

Also: the scoring system grading itself. Buyer clarity 10/10, credibility 9/10. Those are axes that Wishdeal Studio set up and Wishdeal Studio scored. I'm sure there's methodology behind it but calling your own credibility 9 out of 10 is a thing that makes me squint.

## What would convince me

I want to see one person who bought the $5 dossier or the $99 adopt tier, built something from it, and got even 3 paying customers. Not a revenue number. Just confirmation that the doc was actionable and someone executed against it. A forum post, a tweet thread, a link to their product, anything. The "1 in 8 meaningful-success odds" is refreshingly honest but also means 7 out of 8 people who buy this go nowhere. I want to know what the 1 person did differently.

## What I'd ask in an email reply

1. The Year-1 Fermi estimate is negative $22,500. Does that include the cost of buying the dossier, or is it projecting that the market itself won't generate enough revenue to cover a founder's time even if the product ships and sells? Those are very different statements.

2. The feature list mentions "Auto Kill-Switches" and "Team Controls." Is the $99 code starter actually wired to a real API provider (OpenAI, Anthropic) or is it a UI shell I'd need to instrument myself?

3. You score "speed to MVP" at 4 out of 10. What's the blocker? Is it API complexity, getting companies to hand over their API keys, something else?

## Verdict: on-the-fence

The negative Fermi estimate and the "no live customers" disclosure are doing opposite things to me at the same time: one makes me nervous, the other makes me trust them more than I expected to. I'm not ready to put $99 down but I'd probably spend $5 to read the dossier, which is probably exactly what they want.

---
*Memo by skeptic persona, generated 2026-06-13. Studio breaks own self-grading loop.*