Research · Prompt
Funded Idea Scraper
Reads recent funding as a signal for proven venture ideas, clustered by problem space so you can spot markets worth entering before they're crowded.
FUNDED IDEA SCRAPER
# Funded Idea Scraper — build spec
## Context / why this exists
I'm hunting for replicable startup ideas by watching where smart money is flowing. Companies raising Series A have, by definition, cleared the seed stage and been validated by investors as having product-market fit. If I can see which problem spaces are getting funded repeatedly and recently, I can identify proven, replicable business models to build in new markets (the "copycat to a new geography/vertical" play).
Build me a scraper + classifier that continuously surfaces recently-funded companies, groups them by problem space, and lets me spot clusters — not one-off deals.
## Goal (one sentence)Enter your email to unlock the full prompt (+114 more lines) and copy it.
Bonus: turn the data into a visual report
Once the scraper has populated data/atlas_signals.db, paste this second prompt to generate a shareable HTML report straight from your data. No dashboard server required.
VENTURE SIGNAL — VISUAL REPORT
# Visual report from the Series A signal data
## Goal
Read the SQLite database produced by the Series A Signal Scraper (data/atlas_signals.db) and generate a single, self-contained visual report I can share — a static HTML file with charts, no server required.
## Input
- SQLite file: data/atlas_signals.db
- Tables: funding_rounds, companies, problem_clusters, scrape_runs (schema already exists)
- Read-only. Do not modify the database.
## Output
- One file: reports/venture-signal-report.html
- Fully self-contained: inline CSS + a single charting library via CDN (Chart.js). No build step; it opens in any browser.
- Also write reports/venture-signal-report.png (a snapshot of the top section) if headless rendering is available; otherwise skip gracefully.
## Sections (in order)
1. Header — date range covered, total rounds, total capital tracked, and number of active problem clusters. Big, scannable numbers.
2. Momentum leaderboard — top 15 problem clusters by round count in the last 90 days as a horizontal bar chart. Label each bar with cluster name, round count, and total raised.
3. Heating up vs cooling down — clusters ranked by change in round count (last 30 days vs the prior 60). Show the top 5 rising and top 5 falling, each with a trend arrow and percent change.
4. Capital flow — a line or stacked-area chart of weekly capital raised across the covered period, split by stage (series_a vs series_b vs other).
5. Cluster cards — for the top 8 clusters, a card with: label, one-paragraph description, company count, total raised, and up to 6 example companies (name, amount, one-liner, source link).
6. Raw appendix — a compact HTML table of every round (company, stage, amount, date, lead investor, source link) with client-side sort and a stage filter.
## Style
- Dark, technical, editorial. Monospace for labels and numbers; a clean sans for body copy.
- One accent color for emphasis; keep everything else muted grayscale. The data is the visual — no clip-art or stock imagery.
- Desktop-first, but it must not break on mobile.
## Rules
- Derive every number from the database; never invent data. Render null values as "—".
- Use amount_usd for all totals (already normalized in the DB).
- Put the SQL used for each section in an HTML comment directly above that section so I can audit it.
- If a section has no data, render it with an empty-state note rather than omitting it.
## Start by
Printing the row counts per table and the min/max announced_date so we confirm the data range, then generate the report.