Overview
A North American CRM platform specializing in field sales engagement partnered with SourceFuse and AWS to transform how reps interact with their CRM tools. Designed for manufacturers and mobile-first sales teams, the solution aimed to reduce manual effort, improve CRM adoption, and simplify post-meeting updates — all through a voice-enabled GenAI assistant.
The Challenge
Sales reps on the go often delayed CRM updates or skipped them entirely, resulting in lost information, delayed follow-ups, and incomplete pipelines. The leadership team sought a smarter solution that could improve data hygiene without adding friction.
Key goals included:
- Enabling voice-based capture of sales meetings and interactions
- Automatically summarizing and structuring CRM entries
- Designing a scalable backend to support thousands of users across SMB markets
The Solution
SourceFuse delivered a working PoC in just weeks — a serverless, GenAI-powered audio-to-insight engine optimized for mobile CRM.
Scenario Example
After a site visit, a rep records a voice note:
“Met with the distributor. They’re expanding in Q3, need pricing next week, and are evaluating our automation line.”
The system transcribes the note, summarizes key actions, and links it to the CRM account — saving time and improving data quality.
Solution Highlights
- Audio Uploads: Secure voice recordings stored via Amazon S3
- Transcription: High-speed, accurate speech-to-text using Deepgram AI
- Summarization: Natural language insights generated via AWS Bedrock-powered LLMs
- Processing: Lambda + SQS orchestration for async handling
- Scalability: Backend designed to handle 10K+ voice requests per day
The Result
-
Voice-enabled productivity
Reduced manual entry through seamless voice capture and auto-summarization -
80%+ transcription accuracy
Achieved strong results in real-world field tests -
Enhanced rep adoption
Simplified workflows increased engagement across sales teams -
Built for growth
Serverless architecture supports future scale and feature expansion -
Restored delivery momentum
Successfully rebooted after a prior failed implementation
Technologies & AWS Services Used
AWS Services
- Amazon S3
- AWS Lambda
- Amazon API Gateway
- Amazon SQS
- AWS Bedrock
- AWS CloudWatch
- AWS SSM Parameter Store
- Amazon DynamoDB
- AWS CloudTrail
- AWS SES
Technology Highlights
- Deepgram AI for transcription
- AWS Bedrock LLMs for summarization
- Terraform for infrastructure automation
