From what I understand, is that the labels can interfere with the interaction. The his-tag is relative short and the Fc-tag is a big tag. But looking at your data it seems that the Ag.mFc binds better. Did you check for cross-reactivity between the anti-hIgG-Fc and the Ag.mFc? Maybe the cross-reaction is stabilizing the interaction.
One strange thing with your fittings is that the lowest curves in each sensorgram look as if they have a concentration below the KD of the interaction but actually the lowest analyte concentrations is in both cases above the KD.
If the Ag.mFc has two Ag-sites fused to one mFc, than the analyte should be considered bivalent and has consequently a higher affinity due to the two binding sites which stabilize the overall interaction. If there are two mFc-sites fused to one Ag, than the analyte is considered monovalent in this experiment.
In general we like to see that the analyte concentrations are between 0.1 and 10x KD. This is assumed the best concentration range to cover the interaction. With only analyte concentrations below the KD the Rmax is not properly determined and the kinetics are not reliable. With only analyte concentration above the KD there may be a possibility that the calculated kinetics are slightly off (faster association rate).