Vaccination, one of the most effective medical interventions ever introduced, primes the immune system to respond to future infections, dramatically reducing the morbidity and mortality of many infectious diseases. The success of vaccination often depends on the careful selection of adjuvants, immunopotentiating formulations or compounds that elicit a stronger and more durable immune response to the administered antigen. Yet, despite their importance in the development of modern vaccines, few new vaccine adjuvants have entered the clinic in recent decades. In this work, we expand the library of potential vaccine adjuvant candidates by targeting the innate immune pattern recognition receptors nucleotide-binding oligomerization containing protein 2 (NOD2) and Toll-like receptor 7 (TLR7). First, we optimised the structure of a previously reported desmuramylpeptide NOD2 agonist using ligand-based design approaches. We identified key structural features required to achieve immunostimulatory activities in vitro and in vivo. This led to significantly improved adjuvant activity in vivo and the discovery of the first desmuramylpeptide with NOD2 agonist activity in the single digit nanomolar range. Second, we construct a homology model of human NOD2 using a recently reported crystal structure of rabbit NOD2 as the template and apply a hybrid docking/pharmacophore modelling method in search of novel NOD2-modulating scaffolds. Surprisingly, although our structure-based in silico approach did not yield any NOD2 agonists, it did identify two novel NOD2 antagonist scaffolds, which represent valuable new leads suitable for further optimisation. Finally, we extend the adjuvant potential of NOD2 agonists by covalent conjugation with TLR7 agonists. The resulting conjugated co-drugs induced potent Th1/Th17-type T cell immune responses in vitro and Th1 humoral immune responses in vivo. Our study has therefore shown that conjugates can generate robust higher-order interactions that individual pattern recognition receptor agonists simply cannot access.