This work introduces an alternative workflow for the discovery of novel bacterial biocontrol agents in wheat against Fusarium head blight (FHB). Unlike the mass testing of isolate collections, we started from a diverse inoculum by extracting microbiomes from ears of field-grown plants at grain filling stage. Four distinct microbial communities were generated which were exposed to 3 14-day culture-independent experimental enrichments on detached wheat spikes infected with F. graminearum PH1. We found that one bacterial community reduced infection symptoms after 3 cycles, which was chosen to subsequently isolate bacteria through limiting dilution. All 94 isolates were tested in an in vitro and in planta assay, and a selection of 14 isolates was further tested on detached ears. The results seem to indicate that our enrichment approach resulted in bacteria with different modes-of-action in regard to FHB control. Erwinia persicina isolate C3 showed a significant reduction in disease severity (Fv/Fm), and Erwinia persicina C3 and Pseudomonas sp. B3 showed a significant reduction in fungal biomass (cGFP). However, the mycotoxin analysis of both these treatments showed no reduction in DON levels. Nevertheless, Pantoea ananatis H3 and H11 and Erwinia persicina H2 were able to reduce DON concentrations by more than 50%, although these effects were not statistically significant. Lastly, Erwinia persicina H2 also showed a significantly greater glucosylation of DON to the less phytotoxic DON-3G. The bacterial genera isolated through the enrichment cycles have been reported to dominate microbial communities that develop in open habitats, showing strong indications that the isolated bacteria can reduce the infection pressure of F. graminearum on the spike phyllosphere.