Spoken Kitchen — Market Research (powered by Nimble)

Market research for Spoken Kitchen: an AI-powered, bilingual heirloom recipe book for immigrant families. It transcribes and translates the voice of an elderly family cook into an organized bilingual recipe book — available online and as a physical book.

This repo documents how I used Nimble to do the entire go-to-market research stack — ideal customer profile, competitor pricing, positioning, market landscape, and white-space analysis — and synthesized it into a single HTML report for product and marketing decisions.

Why this exists

Spoken Kitchen sits at the intersection of three markets. The research is organized around proving that the gap between them is real and underserved:

1. Emotional / legacy products for parents — e.g. StoryWorth, Remento, Storii
2. Recipe & cookbook creators (online + physical) — e.g. CreateMyCookbook, Heritage Cookbook, Paprika
3. AI transcription + translation apps — e.g. Otter, Whisper-based tools, DeepL

Structure

Path	What's in it
research/01-landscape/	Competitor discovery across all 3 categories
research/02-pricing/	Scraped competitor pricing → comparison matrix
research/03-positioning/	Messaging, value props, marketing strategy teardown
research/04-icp/	Ideal customer profile synthesis
research/05-gaps/	White-space / gap analysis (the thesis)
data/raw/	Raw Nimble responses (JSON + scraped pages) behind every finding
data/processed/	Cleaned CSV / tables
docs/	The final HTML report (index.html, GitHub Pages)
scripts/	Reproducible Nimble commands

The report

The synthesized findings — competitor landscape, pricing, positioning, ICP, and the white-space thesis, with the product & marketing decisions they imply — are in a single, self-contained HTML file with all charts embedded:

• docs/index.html — open it in any browser.

Per-phase detail lives in research/0N-*/findings.md; PLAN.md maps each business question to the Nimble capability that answered it.

Reproducing it

Requires the Nimble CLI with an API key.

export NIMBLE_API_KEY="your_key"     # the CLI reads this from the environment
nimble --version                     # verify install

The CLI reads NIMBLE_API_KEY from the environment, not a .env file. If you keep it in .env, load it first: set -a; . ./.env; set +a. Searches here use --search-depth lite|deep (the fast tier and --include-answer require an enterprise account and otherwise return 403).

Re-run a phase, then rebuild the charts and report:

./scripts/01-landscape.sh            # (and 02–04) → raw output in data/raw/

python3 -m venv .venv && . .venv/bin/activate && pip install pandas matplotlib
python scripts/analysis/pricing_matrix.py   # pricing matrix + chart
python scripts/analysis/gap_map.py          # capability matrix + gap charts
python scripts/build_report.py              # → docs/index.html