In the white noise model we construct an adaptive estimate for f(0) for a decreasing function f. We also show that the maximum mean square error of this estimate attains the same rate as the minimax risk simultaneously over a range of Lipschitz classes of order less than or equal to one.