WinIt automates the quality control of phone calls and SMS
Description
Freja is a company that specializes in creating customizable chatbots for businesses.
Industry
SaaS
Size
11-50 employees
Customer since
Integrations
WinIt and Stack AI partnered to implement an AI system that automates the quality control process of external communications. The system reduced the time spent by the operations team by 90%. WinIt launch the system without expanding its technical team, ensuring the maximum level of data privacy and security.
Opportunities: Performing manual quality check of phone calls and SMS at scale is a time-consuming process
Even with the most sophisticated CRM, operations teams require dozens of individuals to run a manual verification process. This process is not only time-consuming but also error-prone. The operations team needs to transcribe calls and follow a set of rules consistently to ensure the quality of the communication.
- Time: The process consumes a significant amount of time of operations teams.
- Errors: Manual quality control is prone to errors.
- Consistency: Ensuring consistent alignment with company standards is challenging.
- Scope: Only a subset of calls and messages are typically analyzed, given the high volume.
Solution: Implementing an Automated Quality Control system using Audio-To-Text models
WinIt implemented a set of AI internal tools using Stack AI's platform, with the purpose of: a) transcription of phone calls and classify the call based on the quality standards, b) reception of text message, analysis under a set of rules, and classification based on the quality level. The system was implemented using the multimodal capabilities of Stack AI combining DeepGram models, for audio transcription, and Anthropic models hosted in AWS Bedrock (i.e., for maximum data privacy and security), to analyze the call or message.
- AI-Driven Transcription: Using DeepGram models for accurate transcription of phone calls.
- Quality Control: Using Anthropic models on AWS Bedrock to analyze and classify calls and messages based on their quality.
- Batch Processing Interface: Exporting an interface where hundreds of calls and message can be uploaded for quality control, automating the process at scale.
Results: Reduction of the Time Spent and Error Rate of the quality control process
Automating the quality control process allowed WinIt to reduce the time spent on the process by 90% and the error rate by 80%. WinIt scaled its quality control process to a larger volume of phone calls and messages without increasing the number of employees.
- Time Reduction: It took 14 days to analyze a subset of calls. With Stack AI, it took 14 hours. For SMS verification, it took 3 days manually, 3 hours with the AI system.
- Error Rate Decrease: Lowered the error rate in the quality control by 50%.
- Scope Increase: Enabled WinIt to scale their quality control procedures without increasing employee headcount.