Client Background
- Client: A leading pest control firm in the USA
- Industry Type: Pest Control Services / Home Services (Service-Based Industry)
- Products & Services: Home Services & Facility Management, Field Services & Local Services Automation, Consumer Services (Pest Control & Property Maintenance)
- Organization Size: 200+
The Problem
To build a conversational AI bot capable of selling pest control services to live customers via chat, with a future goal of enabling voice interaction. The initial MVP would be a chatbot trained on historical customer call data, integrated with a SaaS platform, and capable of handling sales objections, quotations, and booking flows.
Our Solution
- Designed and developed an AI sales bot using AWS Lex.
- Created backend logic using AWS Lambda functions for dynamic conversational flows.
- Integrated Amazon Transcribe for voice-to-text processing of customer audio calls.
- Added OpenAI’s ChatGPT API via EC2 to improve natural language capabilities where AWS Lex limitations were met.
- Built a pest-identification model (ongoing) to categorize and respond with targeted packages.
- Implemented quotation logic, user data capture, and objection-handling sequences.
- Developed an admin-facing UI (planned) to allow bot customization per company strategy.
Solution Architecture
- AWS Lex – Core conversational interface.
- AWS Lambda – Handles logic, slot fulfillment, response customization.
- Amazon Transcribe – Converts historical audio calls to text for training and insights.
- AWS EC2 – Hosts OpenAI API integration for enhanced NLU responses.
- Python & Flask – Used for web tools, transcription app, and integration flows.
- Future voice integration planned using Amazon Polly + Lex V2 voice support.
Deliverables
- Functional chatbot on AWS Lex (Bot Name: pest_Control)
- Lambda functions for intent handling and slot fulfillment
- Transcription tool for customer call dataset (via Flask + AWS Transcribe)
- Pest classification logic and quote response framework
- EC2-hosted OpenAI integration for advanced language capabilities
Tech Stack
Services: AWS Lex, Lambda, EC2, Transcribe, S3
Languages: Python, Flask
External: OpenAI GPT API
Tools: Amazon Transcribe, OpenAI, Visual Studio Code
Skills Applied:
- Conversational AI Design
- Serverless Architecture (Lambda)
- Natural Language Understanding (NLU)
- Voice Data Preprocessing & Transcription
- SaaS Integration
- Sales Process Mapping
- API Integration (OpenAI, AWS)
Databases:
AWS Buckets
Cloud Server:
AWS
Technical Challenges Faced
- Lex’s limited conversational capabilities and static flow
- Lambda function size and timeout limits when integrating with external APIs like OpenAI
- Difficulty handling voice input and turning it into usable text
- Building a flexible, dynamic quote generation logic based on pest type and customer responses
- Using years of audio recordings as training/reference data
How the Technical Challenges Were Solved
- Integrated OpenAI’s GPT via an EC2 instance to enhance conversational depth
- Offloaded OpenAI API calls to a dedicated EC2 server to bypass Lambda constraints
- Used AWS Transcribe to convert voice input into text and feed it into Lex
- Created a dynamic flow for quoting based on pest category, with follow-up logic
- Developed a custom tool to upload and transcribe audio files for bot training and refinement.
Business Impact
- Reduced need for human reps for first-level queries and quotes
- Created foundation for future voice bot deployment
- Positioned the business for scalable lead conversion automation
- Webapp to simply upload the audio for transcribing.
Project Website URL
