Add a new quality characteristic Vendor Independence to the model.
Modell austauschbar (kein Lock-in) — the model is swappable without prohibitive rewrite cost. The system isolates vendor-specific APIs, capability assumptions, and pricing models behind boundaries so the underlying provider can be changed.
Relevant for AI/LLM-based systems where vendor risk (price changes, deprecations, capability regressions, geopolitical/regulatory shifts) is a real concern. Touches abstraction layers, eval-based qualification of alternative models, and prompt portability.
Suggested mapping
- File: `_qualities/V/vendor-independence.md`
- Tags: `ai`, `flexible`, `portable` (candidates — to confirm during authoring)
- Related: `portability`, `interoperability`, `modifiability`, `vendor-neutrality`
- Standards: ISO/IEC 25010 portability characteristics, open model formats (ONNX, GGUF)
Part of a 9-issue series of AI-specific characteristics proposed by Peter Hruschka. See #447 (Groundedness) and the companion issues for Calibration (#448), Prompt Evolvability (#449), Injection Resistance (#450), Human Controllability (#451), Eval Coverage, Model Transparency, and Drift Detectability.
Suggested by Peter Hruschka.
Add a new quality characteristic Vendor Independence to the model.
Relevant for AI/LLM-based systems where vendor risk (price changes, deprecations, capability regressions, geopolitical/regulatory shifts) is a real concern. Touches abstraction layers, eval-based qualification of alternative models, and prompt portability.
Suggested mapping
Part of a 9-issue series of AI-specific characteristics proposed by Peter Hruschka. See #447 (Groundedness) and the companion issues for Calibration (#448), Prompt Evolvability (#449), Injection Resistance (#450), Human Controllability (#451), Eval Coverage, Model Transparency, and Drift Detectability.
Suggested by Peter Hruschka.