# Dan Kowalski, Senior Software Engineer at FarPoint Logistics (Chicago, 210 employees) — read of AgentOS, June 20, 2026

> 11 years writing backend code for a logistics company, currently plotting my exit by shipping something of my own before I turn 40.

## How I got here

Someone I follow on Twitter (still calling it Twitter) retweeted a post that said something like "we score ideas and publish the math, including when the math is bad." That framing is unusual enough that I clicked. I was on my lunch break eating a sad desk salad. I read the whole page. I went back to work without finishing the salad.

## What I clicked first

"Build a 46-Agent AI Office on Your Laptop." I stared at that line. The 46 is the thing. Not "dozens" or "a fleet" or "unlimited." 46. That specificity either means there is a real architecture behind it or someone in marketing thought a specific number sounds more credible than a round one. I kept reading to find out which. I still do not know.

## Where I paused

The Fermi estimates. I have never seen a product page show me "$-11,500 Year-1 take-home" and "1 in 6 Meaningful-success odds" and then ask me to buy something. That stopped me. My first instinct was that it is a deliberate trust play, designed to stand out from the usual "join 10,000 builders crushing it" homepage. My second instinct was: if you are showing me a negative number and a 17% success rate, what exactly are you selling me? I spent more time than I expected staring at how "buyer clarity: 10/10" and "financial upside: 1/10" can coexist on the same card.

## What I distrusted

A few things. "Build a 46-Agent AI Office on Your Laptop" is a headline that sounds compelling and explains nothing. What is an agent office? What does it do for a customer? Who buys this and what problem does it solve for them? By the end of the page I could not explain what AgentOS actually is to my wife. That is a bad sign.

"We shipped the strategy package; you ship the customer conversations" is doing a lot of work. It means: we made the map, you do the hiking. Fine. But if there are no live customers ("Honest disclosure: we don't have live customers on this idea yet"), I do not know how the map was validated either. That is a circular problem they are gesturing at but not actually solving.

Also, self-scoring "credibility: 9/10" is a strange move. That is not credibility. That is confidence. Those are different things.

## What would convince me

One real case study where someone took a Wishdeal dossier, built the product, and has paying customers. Not a testimonial with a headshot. An actual before-and-after: here is the person, here is what they built, here is what they were making 90 days after buying the dossier. Even small numbers work. $600 MRR from one real operator beats all the Fermi math on the page.

A two-minute screen recording of what an "agent office" looks like when someone is actually using it would also go a long way.

## What I'd ask in an email reply

1. What is the actual product someone builds when they adopt AgentOS? Is there a working prototype somewhere I can look at, or is it a market category description plus a build plan?
2. "Speed to MVP: 4/10" is the concern you flagged. How many hours are we actually talking? What is the realistic estimate for someone with a Rails background and maybe 10 hours a week?
3. Have any of the other ideas in this catalog (Preen, Deltadb, Remedix) been built by real operators, and is there anyone I can talk to about what that experience was like?

## Verdict: on-the-fence

The honesty is rare enough that I did not immediately close the tab, and the $5 unlock is cheap enough that I would probably pop it just to see if the dossier explains what the homepage does not. But I walked away from the free page with zero understanding of what AgentOS actually is as a product, which should not happen.

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