01 / DECIDE
Architecture decision
Use it for document intake, visual QA, accessibility, and mixed text-image analysis.
Preserve the original asset, define allowed observations, and separate visual extraction from consequential decisions. Map input, output, state, tools, and side effects as one observable path, then name the owner of use images as bounded evidence.
02 / OPERATE
Three-step field procedure
- 01
Lock the contract
Write down the caller, data classification, success condition, timeout, cancellation, and ownership. Use it for document intake, visual QA, accessibility, and mixed text-image analysis.
- 02
Run one bounded path
Build one end-to-end path with correlation, typed state, and reversible failure handling. Preserve the original asset, define allowed observations, and separate visual extraction from consequential decisions.
- 03
Verify the signal
Turn acceptance into a repeatable fixture, contract test, or browser test. A labeled set measures extraction accuracy, abstention, OCR errors, crop, and resolution sensitivity.
03 / BOUND
Operational boundary
Do not use visual output alone for identity, diagnosis, safety certification, or disciplinary action.
04 / PROVE
Acceptance evidence
A labeled set measures extraction accuracy, abstention, OCR errors, crop, and resolution sensitivity.
SOURCE / HTTP
Reproducible source probe
curl -fsSI 'https://docs.x.ai/developers/model-capabilities/images/understanding' | sed -n '1,5p'