For all positive $a, b, c$, it holds that $\sqrt{\frac{a}{b+c}}+\sqrt{\frac{b}{c+a}}+\sqrt{\frac{c}{a+b}}>2$. (If we allow nonnegative $a, b, c$ such that at most one of them is $0$, then equality is achieved e.g. when $a=0, b=c=1$.)
This can be proven by assuming $a+b+c=1$ and defining $f(x)=\sqrt{\frac{x}{1-x}}$. Then the sum becomes $f(a)+f(b)+f(c)$. As $f$ has a single inflection point (at $x=\frac{1}{4}$,) we can apply n-1 EV to deduce that if $f(a)+f(b)+f(c)$ achieves a minimal value, then two of $a, b, c$ must be equal to each other. The rest is just calculus.
This method seems convoluted and not particularly elegant. I'm wondering if there exists a proof that doesn't apply any analytic methods (only AM-GM, Cauchy-Schwarz, etc.) If such a proof doesn't exist, what differentiates this inequality from the pretty similar Nesbitt's inequality $\frac{a}{b+c}+\frac{b}{c+a}+\frac{c}{a+b}\ge\frac{3}{2}$ that can be proved, say, with AM-HM? Which class of problems cannot be solved with elementary (olympiad) inequalities and why?