AI automation

Text classification for qualitative research

Themewise · Sort thousands of survey answers into your own themes in seconds, no training needed

WatchingDraft scoreEmerging market · since 2023 · AI estimate

34PWAS score

Confidence

Based on 8 of 9 factors

Overview

AIThis description was written by a model from public signals. The facts behind it are listed under Evidence.

Problem: Non-technical researchers struggle to classify large volumes of participant responses and qualitative data without building custom models or hiring technical help. Existing workflows lack speed, affordability, or usability for this use case.
Solution concept: A tool that enables non-technical researchers to classify large sets of arbitrary text entries against custom themes or topics without prior training data, machine learning expertise, or significant setup, while delivering results in seconds at minimal cost. The tool should handle conceptual matching (e.g., recognizing that "pasta" and "penne" both indicate Italian food mentions) and distinguish between similar concepts.
Who has it: Non-technical researchers and qualitative researchers who need to classify thousands of participant entries or responses.
How they handle it today: Not stated in the source.
Why now: Advances in large language models and zero-shot classification have made it feasible to perform high-confidence text classification without prior training, lowering barriers for non-technical users.
How it could charge: Likely per-classification pricing (charged per entry or per batch classified), per-researcher subscription, or per-project pricing, given the stated requirement for "negligible cost, eg sub 0.1 cents" per classification.

Verdict

Computed from the factor scores below, never written by a model. It decides nothing on its own: only the gate makes an opportunity a build candidate.

Score breakdown

A model sorted each factor into a rubric bucket; code turned buckets into numbers. Unknown factors are left out of the score and lower confidence instead.

Market validation

Confidence 95 · 2 evidence · A single Hacker News discussion with modest engagement (13 points, 20 comments) and an existing competitor (Jev) shows scattered interest, but no evidence of multiple established products or widespread adoption.

Revenue evidence

Confidence 70 · 1 evidence · Evidence describes a target price point ('sub 0.1 cents per classification') rather than actual revenue, paying customers, or proof that researchers will pay.

Growth

Unknown

No evidence shows whether interest in text classification for qualitative research is flat, growing steadily, or accelerating; a single HN discussion does not indicate category-level momentum.

Problem severity

Confidence 30 · Judged from the description · The opportunity describes researchers struggling with speed, affordability, and usability when classifying thousands of entries manually or via expensive technical help—a costly pain that costs time and money.

Economic proximity

Confidence 30 · Judged from the description · The tool directly saves researchers labor and time (and potentially money on hiring), making it a direct cost-saving product that sits near money changing hands.

Distribution

Confidence 30 · Judged from the description · A text classification tool used by individual researchers to classify their own data is a private, internal workflow; the outputs are not inherently shared beyond the user's team.

Buildability

Confidence 30 · Judged from the description · Building a usable text classification interface and integrating a modern language model API (e.g., OpenAI, Claude) is moderate complexity—a standard tech stack plus one or two third-party APIs.

Differentiation

Confidence 70 · 2 evidence · Jev already exists and solves the core problem; the new entrant's edge would be at most a minor angle such as better UX, pricing, or niche customization rather than a structural gap.

Retention and timing

Confidence 30 · Judged from the description · Text classification is occasional use tied to specific research projects, not a daily or embedded workflow; no clear timing tailwind is evident.

Risk penalties

Depends on one platformA viable product likely depends on a single large language model API (OpenAI, Anthropic, or similar) for classification quality, creating dependency on that platform's availability, pricing, and terms of service.−6

Score history

34 · confidence 46 · config v1First score · 8 Oct 2026

Text classification for qualitative research

Overview

Verdict

Build this business

Score breakdown

Risk penalties

Score history

Validation gate

Evidence · 5 items

Market validation

Revenue evidence

Differentiation

Market

Existing products

Build

Legal reconstruction check

Community