*This story was originally published January 2024 in the Gallagher Re Global InsurTech Report for Q4 2023
Tomorrow.io: An Overview
Late-stage growth and a view on exit — Tomorrow.io
Headquartered in Boston, Massachusetts, Tomorrow.io is a weather intelligence and climate adaptation platform helping organizations solve their weather-related challenges. It is powered by next-generation space technology, generative AI, and proprietary modeling capabilities, and its customer list includes JetBlue, Fox Sports, Ford, Uber, and the United States Air Force.
Insurers can use Tomorrow.io for many different purposes. For example, they might use it to alert their customers ahead of hail storms, so people can move their cars off the road, into a garage or under other protection, if they feel at risk. This would reduce hail claims and improve customer loyalty (which may be measured by a net promotor score, or NPS).
Tomorrow.io describes its offering as “weather intelligence”; the translation of global weather forecasts into predictive, actionable insights for decision-makers in any industry or company.
For example, instead of telling Uber that it will rain on Monday — leaving it to the company to figure out how that will impact its business — Tomorrow.io will tell Uber to expect a 20% increase in rider demand from 1-3 p.m. on Monday in San Francisco.
This allows Uber to make more drivers available in advance of the demand spike, in order to reduce ETAs and provide a better customer experience. Based on a company’s specific operating protocols and risk thresholds, Tomorrow.io helps them transition from manual reactive operations to proactive automation at scale.
Tomorrow.io offers many products, all customizable to any use case anywhere in the world. These products include its weather intelligence software platform; an API; historical data; and initial satellite data (following its first two successful satellite launches during 2023).
Enhancing Forecasting With AI & Space
The company enhances its numerical weather prediction models using machine learning; specifically, a post-processing technique that learns from historical forecast errors and observations to correct systematic biases in the models. The system learns patterns in the differences between how the model expects a storm system to evolve, and the observations of what actually occurred. For example, it may identify that there is a warm bias in one area of the models, or that they tend to under-forecast precipitation in one season and/or region. Ultimately, this process produces a forecast that has the lowest error possible over the longest time period possible.
Tomorrow.io launched the world’s first commercial weather radar satellites last year; Tomorrow-R1 and Tomorrow-R2. Equipped
with state-of-the-art precipitation radar, they employ innovative operational modes and meticulous instrument calibration to take precipitation measurements of unprecedented sensitivity and accuracy. While existing satellites update their data every three days, for example, Tomorrow.io’s have an average global revisit rate of one hour. This high-resolution data will lead to a new level of understanding and prediction of hurricanes, floods, landslides, wildfires and droughts. The two ‘Pathfinder’ satellites are thefirst in a constellation that will democratize access to weather data for every person (including the five billion who live outside radar coverage), business, and government on Earth.
The future deployments of additional radar satellites and microwave sounders will enhance their capabilities, recording additional precipitation measurements, atmospheric temperature profiles, water vapor profiles, and offering additional precipitation data proxies. Together, these technologies form the backbone of our cutting- edge weather forecasting system and are expected to enable the collection of multiple geophysical variables.
In 2023, Tomorrow.io also launched the industry’s first weather and climate generative AI. The interactive product, nicknamed Gale, can process vast amounts of weather and climate data, enabling easy identification of key trends, risks, and opportunities. Its key differentiator is Tomorrow.io’s proprietary dataset, from which it draws. Users can specify the weather events relevant to their business operations that they wish to monitor, and Gale can chat about these and/or summarize all of the details into one easy-to-read synopsis.
Tomorrow.io recently released the latest iteration of its weather intelligence platform, known as NextGen. This incorporates
its proprietary satellite data for greater precision modeling, and provides more detailed forecasts than its ‘classic’ product. Tomorrow.io’s full constellation deployment, continuing in early 2024, aims to elevate NextGen’s forecasting precision by an order of magnitude.
While advancements in AI and deep learning are pushing the weather and climate industry into new territories, the datasets behind the models are the core differentiator, with radar data from space being the single most critical component.
Tomorrow.io’s probabilistic suite uses advanced ensemble modeling to quantify the likelihood of different scenarios. Advanced thunderstorm analytics provide hourly storm probability forecasts and detailed severity forecasts to deliver actionable intelligence on tornado, wind, hail, and flood threats up to 14 days out. Together, these capabilities give operators advanced predictive intelligence with hyper-local accuracy to inform high-stakes decisions.
The Power of Weather Intelligence for Insurance
Tomorrow.io has created an approach combining the best of weather observations, prediction and machine learning. It gathers as much data as possible from ground- and space-based observations, and synthesizes it to create customized and hyperlocal forecasts. This unique approach allows the firm to model at the highest spatial and temporal resolutions, while tailoring its predictions and outputs precisely to its customers’ needs — and informing them in detail about the implications for their business.
For insurers, Tomorrow.io’s platform offers:
- Better weather data, enabling value capture across the spectrum
- Improved decision-making before and during extreme weather events
- Improved operational efficiency, with more accurate day-to-day insights and better resource allocation
- Improved risk assessment and more dynamic pricing
- Predictive claims management via the following:
- A historical archive and real-time insights for claim validation
- High-resolution post-event 3D reanalyses
- The ability to notify customers before weather events, to reduce damages/claims
- Targeted customer acquisition and retention
- Dynamic contract pricing
- Improved capital allocation and portfolio optimization
- Product innovation and market differentiation
And it can also support parametric insurance products, by:
- Enabling parametric insurance in Latin America, Africa, India and more
- Giving policy recommendations for specific locations
- Supporting indices with high resolution models
- Supporting business models, with inputs matching product granularity
- Minimizing risk and providing insurance with more certainty
- Avoiding location-based product offerings before/during weather events
“Not only is Tomorrow.io on a literal rocket ship toward improving global forecast capabilities with machine learning being foundational to that, but our weather intelligence and probabilistic forecasting is simplifying weather forecasts into simple predictive operational insights. For instance, as an individual, you might run every morning and your threshold for temperature for wearing a jacket is different than someone else’s. Tomorrow.io allows you to look at the probabilities of a forecast being below a certain threshold where you would want that jacket and make a different decision than somebody else would for high-impact events.”