Technology Requirements for Telco Automation

Telcos and vendors are implementing a steady roll out of automations from task automations towards simple process automations, and then the replication of these simple automations into, say, new channels.   This progression is often a risk-reduction strategy to avoid trying to boil the ocean and limit failure of new projects.  Addition of machine learning is seen, but deployment tends to be limited to information collection exercises and automated root cause analysis – rather than full predictive/prescriptive capabilities.

The diagram above shows a list of new technologies to be added into the next phase of increasingly cross-domain and intelligent automations.  Three newer concepts on the diagram are defined as:

An Automation Map    This holds a summary of various information related to individual automations – including workflows, underlying systems architecture, data in use, points of orchestration and intelligence, teams involved and KPI tracking.  Its main purpose is to provide a single point of insight to allow better management of cross-domain, complex automations.  This includes allowing “big picture” decision making, management of overarching goals, improved risk management and a view of shared resources

Augmented Automation   This is the use of  “augmented analytics” (the use of enabling technologies such as machine learning to assist with data preparation, insight generation and insight explanation to augment how people explore and analyse data in analytics and BI platforms*)  to better manage and understand the data underpinning automations.  The term needs to be broadened within automation to include the understanding of which processes are working well compared to others and the understanding where algorithms/scriptings can be reused to automate similar processes/use cases in a telco

Explainable Automation   An “Explainable AI”** can be easily understood by humans and contrasts with  “black box” AI that employ complex opaque algorithms, where designers cannot explain why the AI arrived at a specific decision.  These capabilities will, presumably, become a necessary part of more intelligent automations as well – and this explain-ability may well need to be ingested into other automations downstream to make them, in turn, explainable.


*Definition of Augmented Analytics from Gartner:

** Definition of Explainable AI from Gartner: