Building the world's first foundation model of the electric grid.
The Hook
“Breaking news. We expect heavy rain falls in India, likely causing floods. Do not step... Deepak, it is sunny and bright outside here. Deepak is not the only one who gets weather forecasts wrong in India.”
Problem Agitation
“This problem costs India thousands of crores because of natural disasters, low agricultural output, and grid imbalances.”
Product Reveal
“This is why we built Indus, the best AI weather model for India, for the electric grid. Over a full year of testing, Indus beats the world's leading forecasts like ECMWF's IFS, their AI model AIFS, and NOAA's GFS on every important grid variable.”
Problem Agitation
“There are two main problems with weather forecasts over India. First, weather forecasts are issued only a few times a day, whereas the grid settles every 15 minutes. And second, these models perform very poorly over India because they're not trained on data from the country.”
Feature Teaser
“We trained a neural net on Indian weather data spanning millions of inputs and learned to characterize hyper-local weather patterns and extreme events, especially ones that other models don't solve, like overpredicting afternoon sun over the Deccan or low wind speed regimes in Western Rajasthan. Every hour, Indus fuses live data from satellites, ground stations, and numerical systems. It issues a fresh forecast every hour, 48 hours out, at 3 kilometers.”
Feature Teaser
“Indus has three time horizons: 48 hours, 10 days, and 7 months. We're working with leading renewable developers in India, and with grid stress increasing every year, accurate forecasts are now a must-have. And many people are already happy with Indus.”
Call to Action
“Mohak is right. I am very happy. Go to weather.pravah.com.”
Synthesizable in Remotion · key components of the grid-forecast-pitch template.
Dynamic numerical matrix where grid cells color-grade from blue to dark red based on RMSE variance against baseline.
Render grid cells using an array map in Remotion. Interpolate cell background color based on numeric values using interpolateColors across lead times.
Geospatial vector field with directional stream arrows animating across regional topology.
Use an SVG overlay displaying vector flow lines; animate strokeDashoffset over frame sequence to simulate moving air currents over India border paths.
Left-docked graphic card sequence appearing with smooth vertical slide and opacity transitions.
Compute frame offsets for each list item using spring physics with damping 12 and mass 0.5 to drive translateY and opacity.
Purpose. Highlight inaccuracy of existing forecasting via recognizable cultural humor.
Execution. Old-school newsroom visual juxtaposed with a clear blue sky street report.
Purpose. Establish monetary and safety stakes for India's electrical grid.
Execution. Rapid news headlines over flood and storm destruction.
Purpose. Prove algorithmic superiority over global weather models.
Execution. Animated RMSE error matrix displaying Pravah Indus outperforming ECMWF and NOAA.
Purpose. Close narrative loop and route users to platform URL.
Execution. Reporters enjoying beach sunshine with persistent URL overlay.