We implemented all the innovations of the year to create third generation of Digital Bot Workers during the last few months.
A Digital Bot Worker (DBW) by definition receives cases / transactions digitally, uses digital process knowledge / guidance, processes the cases digitally and generates transaction deliverables digitally.
The First Generation of Bot Workers were created way back in 2017 and these bot workers had a single ability with an embedded trained neural network model.
The Second Generation of bigger and more capable Bot Workers were created in 2019 and 2020. These second-generation Bot Workers had multiple abilities / skills as compared to the single ability of first-generation Bot Workers. The Second-Generation Bot Workers were also capable of reading one or more SOPs (Standard Operating Process descriptions) provided in the form of a BPMN 2.0 notation. The Second-Generation Bot Workers also were limited to process only English Language documents and content.
The Third Generation of DBWs have just been created with multiple abilities / skills like the previous generation, but with a big ability to understand and use image / data features described using a language called the Patterns Description Language (PDL). These PDL descriptions are created automatically by Teacher Bot Workers. The quality of Bot Worker Decisions also got a boost by the new Reviewer Bot Workers, which use anti-features described by the Teacher Bot Workers. The combination of DBWs, Reviewer Bot Workers (RBW) and Teacher Bot Worker (TBW) provided a on the job learning DBWs. These third gen DBWs also are deployed on a Digital Back Office (DBO) Box which is an enclosure for the DBWs to function. Exciting abilities of the third Gen DBWs are :
1. Much bigger DBWs can be composed with multiple eyes (vision Bots) and multiple Neural Network Models (brains). This means a Single DBW can do multi-process. For example, a DBW with ten eyes can read 10 document pages concurrently.
2. The PDL capability provides on the Job continuous learning and also provides trained DBWs even for rare (sparse sample problem of ML) scenarios.
3. Multi-lingual document and text processing capability. Now these Bots can process English and Arabic Text – printed and handwritten.
4. The Integrated Digital Worker (IDW) capability provides ability to the DBW to login to software systems of the Bank and work like a human worker. The IDW capability is handy when APIs are not available or time consuming to provide.
5. The DBO container can be built with commodity servers and provides a low cost, high reuse, controlled, monitored and audited environment.