One of the questions: "if I use a valuation system of black-box nature, how can I asses model&method risk alongside the valuation?"
OK, if it lets you choose various models for a deal type, you can check on the input/output level.
But discrepancies might lay in the solver of the model? Then you need to check by a system, which tells you how the model solvers, calibration, .. are built and what extended mathematical flexibility means to the accuracy and robustness of your valuation. At least as second source.
Because in risk management, you might need to apply your valuations millions of times. And slight inconsistencies tend to aggregate to unreliable information.
We, at UnRisk, see valuation and risk management as twins. Transparent math is indispensable.
For risk management with less management risk.