# Kavya Krishnamurthy, UX Research Lead at Groww — read of InterviewOS, June 22 2026

> 9 years doing research, currently the only dedicated researcher on a 400-person fintech product team in Bangalore. Run about 30 in-depth interviews a quarter, mostly Hindi and Kannada speakers, mostly on mobile.

## How I got here

I was searching "qualitative research tool Tamil Hindi" after my third week of manually translating interview recordings in Dovetail because their transcription just gives up on code-switched speech. A Substack post from someone at a Bangalore design studio linked to this page. I clicked expecting a live SaaS product. That expectation got complicated fast.

## What I clicked first

The hero stopped me cold: "Preserves cultural nuance and emotional intent that English-first tools miss." That line is accurate in a way that most tool marketing is not. I have a recording right now where a participant said something in Kannada that roughly translates as "it's fine" but the tone and the word choice mean "I've given up trying." Dovetail renders that as fine. So yes, I read on.

## Where I paused

About two-thirds down the page, the framing shifts completely. "Honest disclosure: we don't have live customers on this idea yet. We shipped the strategy package; you ship the customer conversations."

I had to re-read that paragraph twice. So this is not a product. This is a business idea being sold to someone who might build the product. The hero described features, the pricing table showed per-interview billing, I assumed I was evaluating a tool I could sign up for. I was actually reading a pitch deck wrapper for entrepreneurs. The page never flags this transition. That is a significant thing to bury.

## What I distrusted

A few things. The credibility score is 9/10 but there are zero interviews cited, no researcher names attached, no pilot data. The page gives me "thematic clusters, insight patterns, and actionable synthesis" but no screenshot of what that output looks like. What does a synthesis card actually say? What does a thematic cluster look like for a Hindi speaker talking about loan anxiety? The phrase "actionable synthesis" appears in every research tool pitch I have seen since 2021 and it has never once been defined.

Also: the Fermi math is right there on the page, showing year-1 take-home at negative $22,850 and 1-in-8 odds of meaningful success. I respect the transparency but it does make me wonder why anyone would pay $99 to adopt this idea given those numbers. That math is the most honest part of the page and it is also quietly devastating to the pitch.

## What would convince me

If this were a live product: one real output from one real interview. Paste in a raw transcript with code-switching between Hindi and English, show me the synthesis card it produces, let me judge whether it actually preserved what I would have caught manually. That is the whole product claim in one demo. Everything else is decoration.

If this is an idea-for-sale: I would want to know why the team that built this strategy package is not building the product themselves. The stated reason is "you ship the customer conversations," which is founder-voice for "we are not committing." That is a legitimate model but I want it explained directly, not left as subtext.

## What I'd ask in an email reply

1. The page says it understands Hindi, Tamil, Telugu, and Kannada. Is that a trained model on those languages, or is it running a translation step first and then doing English synthesis? Because those are very different products with very different failure modes.
2. Who validated the cultural nuance claim? Was there a researcher involved in testing, or is this a capability assertion based on LLM benchmarks?
3. Is anyone actually building this right now? If I bought the $99 package and started talking to users, would I be the first person doing that or are there two other operators who bought this same dossier already?

## Verdict: on-the-fence

The underlying problem is real and under-served and the language framing is unusually sharp for this kind of page. But I arrived looking for a tool and found a business idea, and the page does not respect that distinction enough to earn my trust. If I knew someone was actively building this and wanted to talk to early researcher users, I would reply. As is I am bookmarking it and waiting to see if a live product materializes.

---
*Memo by skeptic persona, generated 2026-06-22. Studio breaks own self-grading loop.*