Agent 01 · Weather intelligence
Weather.Cal
The first research agent and the data foundation for a wider intelligence network — designed to express what the weather may do, how uncertain that view is and how well it performs by region and horizon.
The purpose
More than a collection of weather stations.
Weather.Cal is being shaped as a regional intelligence producer. The objective is not maximum data volume. It is reliable coverage, known gaps, verified observations and forecasts that can be evaluated where future agents actually need them.
That means treating location, forecast horizon, local conditions and data quality as part of the prediction — not as details to be ignored.
- 01
Regional context
Represent areas and assets with the spatial resolution the downstream problem requires.
- 02
Forecast distributions
Retain ranges and probabilities instead of collapsing uncertainty into one number.
- 03
Ground-truth evaluation
Compare forecasts with observed outcomes and make gaps visible.
- 04
Reusable intelligence
Publish governed outputs that another agent can bind and reproduce.
What the product must communicate
Useful weather intelligence carries its limits with it.
Probability
Distributions and scenario likelihoods rather than unsupported certainty claims.
Time
The exact data cutoff, valid horizon and expected degradation as the forecast reaches further ahead.
Place
Regional meaning that respects geography and does not substitute a convenient station for an entire energy zone.
Quality
Coverage, missing evidence, disagreement and observed calibration remain visible to consumers.
Lineage
The producing agent and exact data-product version stay attached to downstream use.
Reliable means measurably calibrated.
Weather.Cal is not positioned as infallible. Reliability means known error boundaries, transparent uncertainty and abstention when the evidence cannot support a strong view.
The Power.Cal bridge
Weather becomes valuable when the next agent can trust its meaning.
For energy research, Weather.Cal must eventually describe conditions relevant to wind and solar production across real regions and time intervals. Power.Cal will then add generation capacity, demand, grid context and its own uncertainty.
The consumer never inherits a promise of certainty. It inherits a versioned weather distribution and remains responsible for everything it derives from it.
Weather.Cal owns
- Weather variables and spatial meaning
- Forecast uncertainty and horizon
- Weather calibration and evaluation
- Weather-data quality boundaries
Power.Cal adds
- Renewable capacity and generation response
- Demand and system-balance uncertainty
- Grid and market-region effects
- Price-relevant energy scenarios
Current public status
Active research, expanding foundation.
Weather.Cal is validating a narrow first use case while the broader weather-data foundation is designed for regional coverage and future agent consumption. Detailed methods, thresholds, sources and operational telemetry remain inside the protected laboratory.