How natural language processing saves police officers mission-critical time

Tejas Shastry

February 28, 2020


Over the last few years, gk has been developing AI-driven natural language processing software that listens and reads noisy, jargon heavy conversations on Wall Street and extracts important information. Financial firms use gk’s NLP to make better decisions from insight reports and power workflows like form population with your voice.


Recently, we’ve been bringing this technology to public safety institutions that often face a similar problem. Our data science team has been adapting our NLP models for emergency service workers. Police officers can use gk’s models to fill out forms in the field using their voice, eliminating the tedious administrative task of doing them manually at the end of the day. gk’s models easily pick out domain-specific jargon even in noisy environments, so officers can simply describe a situation and gk will take care of the rest. Beyond form population, the gk’s NLP can be used to help power license plate lookups for police officers, reducing the need to “clog the air” with a dispatcher to do the look up instead.


Our data science team was able to adapt our models through a process called transfer learning. We architect our speech recognition and natural language processing such that  a “base model” can learn most of the noise and jargon that officers speak, then the model can be tuned for a particular department. This allows for unparalleled accuracy by making it easy for a department to have a customized model that has learned their officers voices and is deployed entirely within their control. Our focus studio product allows anyone to teach gk’s models new language, jargon, voices, or audio environments.


Learn more about how gk is using natural language processing technology to empower public safety professionals.