AI DevOps & Infrastructure
AWS Migration
Goodcall: Conversation data and semantic search pipelines on AWS
“The Tech 42 team has been instrumental in accelerating our enterprise voice AI initiatives at Goodcall. Their expertise in building robust AWS-based pipelines for conversation data and semantic search helped us turn hundreds of thousands of call recordings into actionable insights. We appreciated their agility and breadth of AI/ML knowledge especially as the scope evolved and their proactive approach ensured our goals were met without losing sight of user needs. This project wouldn’t be where it is today without Tech 42’s support."
Bob Summers
Founder and CEO
Project summary
Goodcall processes hundreds of thousands of call recordings but needed a way to extract insights and train AI models at scale. Tech 42 built a transcription and analysis pipeline on AWS. Lambda orchestrates the workflow. Whisper V3 transcribes audio files stored in S3. Bedrock runs parallel analysis jobs using Claude Sonnet, Amazon Nova, and Llama to generate summaries and extract topics. Results are stored in DynamoDB for semantic search and analytical querying. CloudWatch provides monitoring and visibility. The project includes a model comparison report showing which AI models perform best for summaries and topic extraction. The pipeline turns unstructured call data into actionable insights. Goodcall can now automate call analysis, improve AI agent performance, and drive data-driven customer interaction enhancements.
