Guide: Preparing a care call flow
Long readRunning calls
What has to be decided before the call opens — what gets asked first, how long to wait, and where a person has to take over.
Keep readingGuidesRead the docs
Running care calls
How the calls actually get run
Each guide assumes what exists today. Where a path is not open yet it says so, and the only figures carried here are the ones counted in the repository.
Running calls
What has to be decided before the call opens — what gets asked first, how long to wait, and where a person has to take over.
Keep reading
4 / 4
Guide: Reviewing call transcripts
Medium readOperator console
How to read back what accumulated during a call — what to look for, what to record, and what to carry into the next one.
Keep readingGuide: Staging capability behind flags
Medium readPlatform
A feature flag is a level, not a switch. Risky paths default off and only come up once what they do is observable.
Keep readingGuide: Handling a failed call
Medium readReliability
Failure is chosen, not eliminated. Tell apart what wavered, step down instead of dropping, and close out what is left in the console.
Keep readingThe surfaces the guides run on
Publish history — changes land as versions, and any version can be rolled back. 
Stack settings — model and router picks apply from the next call.
The principles, briefly
Where would you like to start?
Begin with how the stack is put together, or start from the questions people actually ask.
Figures are counted in the repository — audit, August 2026.