AT&T receives around 40 million customer service calls each year, often requiring categorization of various issues to enhance customer retention. Initially, the company utilized ChatGPT for sorting calls, noting its effectiveness but also its high costs and time constraints. Due to these challenges, AT&T transitioned to developing a more economical and faster open-source AI system to efficiently categorize call summaries. The goal is to manage customer issues proactively, preventing churn effectively while streamlining operational tasks.
The GPT-4 model did produce very high-quality outputs... but it was very expensive; sorting the calls was a daily task unsuitable for time-intensive processes.
With large language models, AI can now ingest summaries and categorize calls, preventing customer churn while making the process more efficient.
Collection
[
|
...
]