Sinopah wet from the river, sitting in tall grass with forested banks behind her

Plain language

Logging with an agent

Start here after a hard night, after an idiopathic shrug, or when you want help that can remember your dog from week to week.

How does a caregiver thinking-partner agent differ from a seizure app?

A typical diary or seizure app wants consistent fields and reports. Useful. Limited when your understanding changes midstream. A Ball Throwers agent is a thinking partner on private MCP: it holds one active hypothesis, takes session notes in your language, tallies support and contradict, queues next tests, and returns overviews. Chat stays in Claude or Cursor. The household database stays yours.

Method, not cure. Logging does not replace veterinary care. Patterns you find are questions for the clinic, not permission to change medications alone.

Why did rigid trackers fail some of us?

When the living question changes, fixed schemas punish you. Data entry gets heavy. Analytics lag. There is no room to be wrong out loud and try again. Sinopah's household lived that failure with a 2024 tracker app, then moved the same care problem into an agent that could follow the hypothesis. Your household may need different fields. The product shape is the follow, not a cloned thermal model.

What should I log if I am hunting triggers?

Log for the hypothesis:

Keep love-moves labeled as love-moves. Keep medication changes inside veterinary decisions, not inside "the app said."

How private is this?

Alpha access is a household key. Tools fail closed without it. Session and event detail are not a public feed. Story sharing, if you ever want it, is opt-in and curated. See Start and The Agent.

Method short form: Method. Hypothesis buckets: Hypothesis method.

Back to Resources.