# Dana Kowalski, Founder at Kowalski Digital — read of voicemail-summary-ai, May 13, 2026

> 8 years running a bookkeeping firm for contractors and home-service businesses, sold it in 2023, now trying to build something productized in the same world I came from.

## How I got here
Someone in the Tiny Empires Slack dropped a link to Wishdeal with a "has anyone bought one of these?" and I clicked around for 10 minutes. This one stopped me because I literally watched my old clients' front desks drown in voicemails for years. I'd actually Googled "voicemail AI for contractors" a few weeks earlier and found nothing that felt like a real product. So this felt timely enough to dig in.

## What I clicked first
The hero pulled me in before I had a chance to be skeptical. "Your voicemail backlog just got cleared." That is the right language for this buyer. I spent years watching plumbers and HVAC owners miss a Friday callback and lose the job to whoever picked up the phone first. "Hot leads cool down while the team triages a hundred other messages" -- I've watched that happen in real life, in a real office, multiple times. The pain framing is accurate.

The 30-second explainer link is at the very top and I would have clicked it immediately. If that video doesn't deliver in the first five seconds I'm gone.

## Where I paused
The Wishdeal scoring table. "69/100 Adoptability. $-17,870 Year-1 take-home (Fermi). 1 in 8 Meaningful-success odds."

I had to sit with that. A studio publishing its own negative projected take-home on a product it's selling the blueprint for is either the most honest thing I've seen in this genre or a very clever framing trick. I kept reading to figure out which.

The strongest-axes breakdown helped. "credibility: 10/10, uniqueness: 9/10, buyer clarity: 8/10." OK so they think the concept is solid. The weakness is "financial upside: 2/10." A $92K mid-case ARR with $26K build cost and a negative year-one result is lifestyle-business math, not venture math. That's actually useful. It resets expectations in a way most idea-marketplace pages refuse to do.

## What I distrusted
"near-human accuracy (not the garbled stuff from traditional VM systems)" is doing a lot of work with zero numbers behind it. What is the word error rate? In what environments? Compared to what baseline? Every transcription product says something like this. Show me a clip from a noisy job-site call transcribed next to Google STT on the same audio.

"Transcripts ready before the caller hangs up." I clocked that. Real-time transcription before the call ends is a latency claim that matters in practice. I would need to know what engine is underneath this (Whisper, Deepgram, something else) before I would stake a client's callback pipeline on it.

The bigger trust problem: the feature list describes a finished product -- CRM sync to Salesforce, bulk callback scheduling, smart segmentation -- and then the page admits "we don't have live customers on this idea yet." Which is it? A shipping feature set or a roadmap? That gap is the most friction I hit on the whole page.

## What would convince me
One real voicemail, transcribed, with the summary and action items shown side by side. Not a studio recording. A genuine missed call from a stressed-out customer, with background noise, partial sentences, a phone number repeated twice. If the output is actually readable and useful, I'm sold on the core premise.

On the business side: one real conversation (anonymized is fine) with a contractor or service business owner about whether they would pay $79 or $199 per month, and what would make them cancel. Not a testimonial. An honest "here is what they told us about pricing" note from an actual discovery call.

## What I'd ask in an email reply
1. The $-17,870 year-one take-home assumes what cost structure? I have an existing contact list of about 120 service businesses from my old bookkeeping firm. Does that change the Fermi meaningfully or is it still a year-one loss regardless?
2. The feature list shows CRM sync for Salesforce, HubSpot, and Pipedrive. Is that in the $99 code starter or is it on the roadmap?
3. What is the actual STT engine? Not asking for trade secrets. I want to know if I am duct-taping Whisper together myself or if there is a Deepgram contract already baked in.

## Verdict: curious-enough-to-reply
The self-scoring transparency broke my skepticism enough to keep reading past the fold. Most idea-marketplace pages pretend the downside is low; this one published a negative year-one projection and a 1-in-8 success rate right on the page, and I appreciated that. The pain is real and I have watched it happen with my own clients. The $5 dossier is a low enough bet that I would probably pull the trigger on that before sending the email, but I might do both.

---
*Memo by skeptic persona, generated 2026-05-13. Studio breaks own self-grading loop.*