# Devin Okafor, Marketing Analytics Manager at Nuvelo Health (122 employees) — read of Natural Language Data Workspace, June 9 2026

> 7 years in B2B analytics, currently the person who begs our data engineer for one-off BigQuery reports and feels bad about it every time.

## How I got here

Someone in the Analytics Engineers Slack dropped a link to this with the comment "interesting NL-to-SQL thing, thoughts?" I was on the 7:14 train and had nothing else to read. I typed the URL in manually because I don't click random links from Slack when I'm half awake.

## What I clicked first

The hero. "Query your database in plain English. No SQL required." That's exactly my itch, and I've been burned by it before. Metabase's natural language feature. ThoughtSpot's search bar. The OpenAI-wrapped thing a contractor built us that hallucinated joins. So I was already in skeptic mode before the page finished loading.

The live demo thing caught my eye. The words "Try it Live" promised I could poke at something real. I could not find an actual interactive demo on the page. Maybe it's there and I missed it. If so, the UX failed me.

## Where I paused

The scoring section. I actually stopped scrolling.

"66/100 Adoptability. $-25,680 Year-1 take-home (Fermi). 1 in 9 Meaningful-success odds (Fermi)."

I had to re-read the page from the top. Because I thought I was reading a product homepage for a tool I could subscribe to. Turns out this is something else entirely. This is Wishdeal Studio selling me a strategy dossier FOR building this product, not selling me the product itself.

That realization happened at the bottom of the page. The top half is written entirely in the voice of a product targeting data people who can't write SQL. The bottom half is written for a founder evaluating whether to build that product. These are two completely different readers and the page serves neither of them cleanly.

"Honest disclosure: we don't have live customers on this idea yet. We shipped the strategy package; you ship the customer conversations."

So the product in the feature list above does not exist as a shippable thing. It's a concept. That's a significant gap between what the hero implied and what this disclosure says.

## What I distrusted

The features list reads like someone enumerated every checkbox that NL-to-SQL tools should have and called it a product: "Query Learning," "Audit Trail," "Multi-Database," "Collaborative Workspaces." These are the right things to say. They are also exactly the right things to say if you've never actually built the system and are pattern-matching from competitors. There's no specificity anywhere. How does "the system learns your schema"? Does it read DDL? Does it do RAG over your table names? Does it require me to annotate anything? The feature is stated as if it already works rather than as a capability the dossier is proposing to include.

"Uniqueness: 9/10" is the score I distrusted most. There are at least eight funded companies doing this right now. Some are well past Series A. Uniqueness is the axis where I'd push back hardest before buying a strategy package built around it.

## What would convince me

If this page is aimed at someone considering building this: I'd want to see what the dossier actually covers. One sample page. The competitive map. Which customer segment the Fermi estimate assumes. The "financial upside: 2/10" score is alarming and I'd want to understand the reasoning, not just the number.

If this page is aimed at someone like me who wants to use the tool: show me a 90-second screen recording of someone asking a genuine messy question against a real-looking schema and getting a correct answer. Not a polished marketing query. Something like "show me churned accounts that had more than 3 support tickets in the 30 days before they canceled." Watch the SQL it generates. Show me it's right.

## What I'd ask in an email reply

1. The "1 in 9 meaningful-success odds" - what's the base rate you're comparing against, and what counts as meaningful success? Is that a 10x return on dossier cost or something bigger?

2. Who is the intended buyer of this page right now? Because I read the top half as a user and the bottom half as a founder, and I genuinely don't know which one you're writing for.

3. The multi-database claim ("PostgreSQL, MySQL, Snowflake, BigQuery") - is that in the strategy as a launch scope, or is that the eventual roadmap? Because if you're suggesting someone launch against all four simultaneously, that's a concerning signal about scope judgment.

## Verdict: on-the-fence

The honesty about "we don't have live customers" and publishing a 2/10 financial upside score is genuinely unusual and I respect it. But the page creates two contradictory experiences: I arrived expecting a tool I could try, and I left unsure whether I was the right audience at all. That's a fixable problem, but right now it's costing you replies from both groups.

---
*Memo by skeptic persona, generated 2026-06-09. Studio breaks own self-grading loop.*