All activity
WorldSim runs Monte Carlo simulations across socio-economic indicators for 195 countries, producing full probability distributions (P10/P50/P90) instead of single-point forecasts. Over 100 structural coupling rules connect variables causally: shock energy prices and inflation, rates, housing, and debt cascade automatically. Three scenario paths (Better/Average/Shock), user-controlled bias overlays, and reproducible runs. Built for macro analysts, policy teams, and country risk professionals.

WorldSimProbability distributions for 195 countries to 2050
Yiannis Pitsillidesleft a comment
After years of building ML models and forecasting pipelines, I came to a conclusion that changed how I think about the future: single-point forecasts are almost meaningless for structural questions. When you're modelling what happens to a country over 20 years, one number tells you nothing. What matters is the distribution. The range of plausible outcomes. The tail risks. Over the past year,...

WorldSimProbability distributions for 195 countries to 2050
Yiannis Pitsillidesleft a comment
Hey Product Hunt I built VideoSenseAI after working with long videos (skiing footage, drone videos, recordings) and realizing how painful it is to find specific moments. Scrubbing timelines works — but it doesn’t scale. So I focused on a different approach: turning video into searchable data by indexing visuals and audio together. Instead of watching everything, you can search objects, words,...

VideoSenseAITurn video into searchable data
VideoSenseAI turns raw video into searchable data by analyzing both visuals and audio.
Upload a video (or link) and automatically:
• detect objects across frames and aggregate them
• generate searchable transcripts from audio
• build timelines you can search and jump through
• dashboard-like analytics
• export structured outputs like tables and CSVs
It’s built for anyone working with long videos who wants to find moments, patterns, and insights — without watching everything end to end.

VideoSenseAITurn video into searchable data
