AutoSales Assistant

• 

  Architecture Diagram

Inspiration

Sales teams spend countless hours on repetitive tasks - transcribing meetings, writing follow-up documents, creating technical diagrams, and sending emails. We wanted to build an AI assistant that could automate this entire workflow, letting sales professionals focus on what they do best: building relationships and closing deals.

What it does

AutoSales Assistant is an intelligent chatbot that transforms sales workflows through automation: Transcribes meeting videos from S3 storage with high accuracy Generates professional Minutes of Meeting (MOM) with action items and speaker identification Creates comprehensive Statements of Work (SOW) from meeting requirements Designs system architecture diagrams automatically from technical discussions Sends polished documents via email to stakeholders Researches market trends and competitor information in real-time

##How we built it

Backend: Python with AWS Bedrock for AI processing and Strands Agents framework AI Models: Claude 3 Sonnet for intelligent conversation and document generation Storage: AWS S3 for video/document storage, Memory DB for conversation persistence Frontend: Dual interface with React.js and Streamlit for different user preferences Infrastructure: AWS Lambda for serverless processing, API Gateway for REST endpoints Tools Integration: Gmail API, DuckDuckGo search, Graphviz for diagram generation

Challenges we ran into

Memory Management: Maintaining conversation context across multiple AI tools and sessions Video Processing: Handling large video files efficiently while maintaining transcription accuracy Tool Coordination: Ensuring seamless data flow between transcript → MOM → SOW → architecture tools Real-time Processing: Balancing response speed with comprehensive document generation Security: Implementing secure S3 access while maintaining user-friendly experience

Accomplishments that we're proud of

End-to-End Automation: Complete sales workflow from video input to final deliverables Intelligent Tool Orchestration: AI agent that knows which tool to use and when Professional Output Quality: Generated documents that are actually business-ready Dual Frontend Support: Flexibility for both technical and non-technical users Persistent Memory: Conversations that build context over time for better results

What we learned

AI Agent Architecture: How to design modular AI systems that work together seamlessly AWS Bedrock Integration: Leveraging cloud AI services for production-scale applications User Experience Design: Balancing powerful features with intuitive interfaces Memory-Driven AI: The importance of persistent context in multi-step workflows Hackathon Execution: Rapid prototyping while maintaining code quality and documentation

What's next for AutoSales Assistant

CRM Integration: Connect with Salesforce, HubSpot, and other sales platforms Advanced Analytics: Sales performance insights and trend analysis Multi-Language Support: Global sales team collaboration Voice Interface: Real-time meeting assistance during calls Custom Templates: Industry-specific document templates and workflows Mobile App: On-the-go access for field sales teams Enterprise Security: Enhanced compliance and data governance features

Built With

• amazon-web-services
• claude
• python
• react
• smtp
• strands
• streamlit