We are in the business of bringing efficiency to the Restaurant ordering function. Every milli-second we eliminate from the process of serving customers, helps the Brand save crucial time with cars running in idling, spewing gas continuously and thus causing harmful emission to the planet.
At HUEX we are working on several workstreams to reduce our overall impact to the planet.
- DATA COLLECTION: Our data collection required to build custom model for a brand is minimal. We only need 100 hours or less to achieve high accuracy – Less data collection, means less power consumption for processing & training.
- MODEL TRAINING: When we collect data, there are long pauses and silences in the interaction in between the turn taking between different audience. We eliminate these long periods of silences so we do not have to store audio files that have these gaps resulting in bigger storage space and unnecessary compute / processing resources.
- MODEL INFERENCE: We build our models on the Cloud, but our inference is at the Edge in a very low compute resource environment. Very minimal resource footprint for our client operations.
- PERFORMANCE: Our performance of automated order taking is in the order of ~ for every 1 second of spoken language, the machine is able to understand and generate appropriate response in under 1/2 second. Essentially we are reducing the overall time required by the customer at the drive-thru lane to serve their needs.
At HUEX we are building applications, that use the right compute resources and provide the best value for the end users while being planet friendly.
