Developing antibodies is complex and resource-intensive, and methods for designing antibodies targeting specific epitopes are lacking. We introduce ade novoantibody design approach leveraging the empirical force field FoldX to design complementarity determining regions (CDRs). Starting from a scaffold VHH, we tackled three challenges of increasing difficulty: 1) design the CDRs to optimize VHH stability and affinity for its original target; 2) design the CDRs for high affinity to the human ortholog; 3) design the CDRs for low nanomolar affinity for a pre-defined epitope on the unrelated human Interleukin-9 receptor alpha, for which no antibodies were previously developed. For each challenge we reached single digit nanomolar affinity in a single design cycle. Our approach allowsde novodesign of high-affinity VHHs while ensuring specificity and stability.