Yes, the estimated path coefficient linking one variable to another can inform you as to whether there is a non-zero "influence" or, more neutrally, relationship (of course, you would refer to the test of significance for the coefficient, or to the confidence interval estimate).
As well, if the coefficient reported is standardized (relative to all others in the model), you can interpret the relative impact of that relationship, in comparison to others. Note that AMOS does not, by default, issue standardized coefficients--you have to ask for them explicitly.
Of course, gauging the strength of the relationship of some IV on some DV also must be tempered by the other aspects of the model (e.g., are you evaluating direct vs. indirect effects, or of moderation, or of a set of IVs on a DV--in which case collinearity can cause you to misrepresent absolute relationship strength). So, model complexity matters.