Use case

AI and ML Based Machine Scoring and Quality Management

SuccessKPI enables us to achieve extraordinary levels of transcription accuracy. Their machine learning and automation tools allow us to deploy applications and POCs in a fraction of the time required by previous platforms. 


Company Profile

Comunikime is a global systems integrator responsible for helping large organizations operate IVR, ACD, RPA, and telecommunications platforms at scale. Operating in Latin American, their Portuguese and Spanish-speaking clients have struggled with low transcription success of legacy applications.

Business Situation Overview

Operating in Latin America, Comunikime has enterprise clients with tri-lingual applications (Portuguese, Spanish, and English). Further, their Portuguese and Spanish-speaking populations have struggled with low transcription success of legacy speech technology approaches. The movement to the cloud has enabled Comunikime to integrate tools such as SuccessKPI to leapfrog previous initiates.


SuccessKPI deployed a cloud-based SaaS application in the Amazon cloud, which enabled access to this platform from new cloud deployments on platforms such as Genesys Cloud and with dozens of legacy ACD applications such as on the Aspect Software platform in one unified operating environment. Today, Comunikime can activate analytics and automation projects in the cloud to analyze customer data, enable AI and Machine Learning tools, deliver POCs, and deploy solutions to crucial business performance projects.


Agents are required to adhere to multiple complex processes and to demonstrate skills to manage high net worth clients. With poor transcription rates from a previous platform, clients were required to score calls manually with quality managers.


SuccessKPI activated a cloud-based environment and carried out a dozen client projects providing topic detection, theme detection, machine scoring, integrated analytics, and performance management presentations. This solution achieved nearly 90% transcription accuracy, brought new sentiment analytics, and advanced topic detention and correlation. Finally, the quality management tool enabled calibration of results with manual scores completed by supervisors operating in the evaluations area of the SuccessKPI tool. Machine learning technology learned from human scores to provide score performance comparable to humans.


Comunikime was able to measure performance across 100% of calls at one percent of the previous cost of operation and gained new insights from the robust analytics layer.