At Myrtle.ai we specialize in delivering the highest throughput for DNN inference within strict latency bounds. We are focused on two main areas, Recurrent Neural Networks and Recommendation Systems. Recurrent Neural Networks. We deliver industry-leading performance for applications that contain RNN layers such as speech transcription, machine translation, speech synthesis and natural language processing. If you run inference within strict latency bounds then we can help you achieve more in the time available. The extremely low latency is critical in some applications such as time series analysis and high frequency trading; it also greatly simplifies the scheduling of multiple asynchronous workloads with differing Service Level Agreements. Recommendation Systems. High throughput, low latency DNNs are an imperative for recommendation engines. We have a scalable solution that delivers a 10x improvement in raw bandwidth within existing infrastructure under tight latency constraints. Furthermore, the 78% reduction in energy requirements achievable using Myrtle.ai technology dramatically reduces energy costs and enables major corporations to meet their carbon footprint goals.