Purpose and proportionality
What decision will an advisory help someone make? Which signals are essential? Avoid widening collection simply because another source is available.
RAIN / Resources
A working set of questions for deciding what an on-site AI system should observe, explain and leave to people.
Working product framework
01 / THE IDEA
RAIN begins with operational telemetry and a specific question about an asset. This framework is a guide for scoping that work: understand the purpose, the people affected and the evidence needed to justify the next step.
What decision will an advisory help someone make? Which signals are essential? Avoid widening collection simply because another source is available.
Who can approve access, challenge an output or stop a trial? Keep those responsibilities explicit and keep equipment authority with the operator.
What is known, what is inferred and what has not been tested? Preserve that distinction in model evaluation, product language and pilot reporting.
02 / IN PRACTICE
Define the scope, preserve the boundary, and make the outcome something a person can inspect.
Describe the asset, the question and the people who will use the output. State which decisions the system is not authorized to make.
Consider what happens when an advisory is wrong, late or misunderstood. Plan an operator review process suited to that consequence.
Document the agreed scope, open questions and reasons to pause or revise the trial. Revisit them when conditions change.
03 / A FEW DETAILS
No. It is RAIN’s working product framework for discussing intended use and responsibility. It does not establish certification or regulatory compliance.
No. RAIN’s current evidence comes from a battery-storage simulator. Other applications need their own integration work, use boundaries and validation.
BUILD WITH RAIN
RAIN is preparing for its first field pilot. Start by defining the environment, the data boundary, and what success should look like.