Hamilton Perkins Collection
Hamilton Perkins Collection: AI Chatbot for B2B Recycling Quotes and B2C E-commerce
Industry
Sustainability / E-commerce / B2B RecyclingProject type
AI Chatbot Development / Web Application IntegrationProject duration
2024 - In Progress (6+ Months)Team size
5 specialists
At a glance
Project milestones delivered:
B2B Quote Automation
Reducing quote generation from days to under 60 secondsDual-Purpose Intelligence
Single AI chatbot serving both B2B and B2C audiences24/7 Availability
Automated quote generation and customer support around the clockCore Functions
Automated B2B Recycling Cost Estimates & B2C Sales AssistancePlatform
AI-Powered Conversational Interface
About Hamilton Perkins Collection
Hamilton Perkins Collection (HPC) makes premium bags and accessories from recycled materials: plastic bottles and reclaimed billboard vinyl. They run two distinct businesses under one brand. A direct-to-consumer e-commerce operation selling to environmentally conscious shoppers, and a B2B recycling operation that sources and processes materials for manufacturing.
As HPC scaled, they hit a wall: two audiences with completely different needs, both requiring immediate, knowledgeable support. Manual processes couldn't keep up.
Founded
Headquarters
Market
The Challenge
The problems
B2B Quote Bottleneck
Every recycling inquiry required manual processing through email chains and spreadsheet calculations. Sales teams spent 2-3 days per quote gathering material type, quantity, location, and logistics information. That delayed partner onboarding and pulled resources away from closing deals.B2C Cart Abandonment
E-commerce customers had questions about sustainability credentials, material composition, care instructions, and shipping. Without immediate answers during browsing, potential buyers abandoned carts.Resource Strain
A small team dedicating specialists to repetitive quote requests and basic customer inquiries meant fewer people working on growth, product development, and partner relationships.Inconsistent Information
Multiple team members handling inquiries across B2B and B2C channels. Pricing details, product specifications, and sustainability data sometimes conflicted between channels.
Constraints
Brand Voice Consistency
The chatbot needed to match HPC's authentic, mission-driven brand voice. Generic chatbot responses would undermine customer trust.Technical Integration
Integration with existing e-commerce platforms, inventory systems, and CRM tools without disrupting operations.Data Security
Enterprise-grade security for sensitive B2B business information shared during quote requests.
The goal
Build an AI chatbot that generates accurate B2B recycling quotes in under 60 seconds while simultaneously serving as a B2C shopping assistant. Both audiences served 24/7 from one interface.
Our solution
Why HyperSense
They needed a team that could build conversational AI handling complex business logic while maintaining natural, brand-aligned interactions. Our experience with AWS Bedrock, Anthropic Claude, and building AI systems for dual-audience use cases made us the right fit.
Our approach:
We started with deep discovery across HPC's sales, operations, and customer service teams. Analyzed historical customer service transcripts and B2B email threads to understand how people actually communicate about recycling quotes and products.
We used an iterative MVP approach. B2B quote automation launched first to validate the calculation logic and data collection flow. B2C capabilities layered in using insights from initial deployment. This let HPC see value immediately while we refined the conversational experience from real interactions.
What we delivered
Core platform capabilities:
Intelligent conversation orchestration
Distinguishing between B2B quote requests and B2C shopping inquiries based on user intentB2B quote engine
Structured data collection workflow for B2B recycling quotes (material type, quantity, location, specifications)B2C product knowledge
Natural language product knowledge base for B2C customers with sustainability impact informationSeamless human handoff
Seamless human handoff when complex inquiries exceed automated capabilities
Supporting Infrastructure:
Express.js API backend
Integrating quote calculation logic with inventory and pricing systemsPostgreSQL database
Storing conversation history, quote records, and customer interaction analyticsReact chat widget
React-based chat interface widget embedded in HPC website with responsive Tailwind CSS designAWS Bedrock Data Automation
Continuous knowledge base updates from product catalog changes
Integration & Deployment:
Website integration
Website integration with minimal impact on existing e-commerce platform performanceCRM synchronization
CRM synchronization capturing qualified B2B leads and B2C customer insightsAnalytics dashboard
Analytics dashboard for conversation patterns, quote volume, and conversion metrics
Architecture & technology
AWS Bedrock's managed AI infrastructure handles the heavy lifting. No custom model training required. The serverless approach scales during traffic spikes without paying for idle capacity.
AI foundation
Technologies
AWS Bedrock, Anthropic Claude
Purpose
Natural language understanding, context-aware conversation management, intent recognition
Data automation
Technologies
AWS Bedrock Data Automation
Purpose
Continuous knowledge base sync from product catalogs, pricing updates, and business rule changes
Frontend widget
Technologies
React, Tailwind CSS
Purpose
Chat interface embedded in HPC website, brand-consistent styling, mobile-optimized
Backend API
Technologies
Express.js, Node.js
Purpose
Quote calculation engine, conversation routing, CRM integration, analytics
Database
Technologies
PostgreSQL
Purpose
Conversation logs, quote history, interaction analytics
Cloud infrastructure
Technologies
AWS (Lambda, API Gateway, S3, RDS)
Purpose
Serverless computing, API management, managed database hosting
Technical highlights
Anthropic Claude's extended context window enables multi-turn conversations where the chatbot remembers previous statements. Critical for B2B quote workflows where users provide information incrementally. The system dynamically adjusts conversational style based on detected intent: professional, specification-focused dialogue for B2B partners and friendly, benefit-driven language for B2C shoppers.
Sub-500ms response times for typical interactions through strategic caching of knowledge base content and pre-computed quote calculation templates.
Key Features & Capabilities
Automated B2B Recycling Quote Engine
What it does:
The chatbot guides B2B partners through a structured conversation collecting material type, quantity, pickup location, and logistical requirements. Generates accurate cost estimates in under 60 seconds.Why it matters:
This turns a multi-day manual process into an immediate interaction. Partners get instant quotes without waiting for sales team availability. The system captures and qualifies leads 24/7 across all time zones.Technical approach:
Quote calculation engine integrates real-time pricing data, shipping cost APIs, and business rules defined by HPC's operations team. AWS Bedrock Data Automation ensures pricing updates propagate to the knowledge base within minutes.
B2C E-commerce Shopping Assistant
What it does:
For retail customers browsing the HPC website, the chatbot answers questions about materials, sustainability impact, product care, sizing, shipping, and returns. It suggests products based on preferences expressed during conversation and guides shoppers toward purchase decisions.Why it matters:
Immediate, accurate answers at the moment of need. Reduces cart abandonment by addressing common friction points: product uncertainty, lack of support during browsing, and hesitation about purchasing decisions.Technical approach:
Knowledge base draws from HPC's product catalog, FAQ documentation, sustainability certifications, and customer service history. Natural language understanding identifies implicit concerns, like durability questions signaling hesitation about recycled materials, and proactively addresses them.
Intelligent Conversation Routing
What it does:
The system detects whether a user is a B2B partner or B2C customer from conversational cues, question patterns, and contextual signals. No "Are you a business or consumer?" prompt needed. Language, tone, and information depth adapt automatically.Why it matters:
One platform to manage, one knowledge base to update, one analytics dashboard to monitor. Two specialized experiences for two distinct audiences.
Seamless Human Escalation
What it does:
When conversations exceed the chatbot's capabilities, the system transitions to human team members with full conversation history and context preserved. No information repeated. Specialists immediately provide high-value assistance.Why it matters:
Escalation analytics identify knowledge gaps and conversation patterns where expanding automated capabilities would add the most value.
Implementation journey
Discovery & Architecture Design (4 weeks)
- Conducted comprehensive stakeholder interviews with HPC sales, operations, customer service, and leadership teams to understand workflows, pain points, and success criteria
- Analyzed historical customer service transcripts and B2B email threads to identify common conversation patterns, information requirements, and decision points
- Designed conversation flow architectures for both B2B quote workflows and B2C shopping assistance scenarios with clear escalation paths
- Established brand voice guidelines and conversational tone parameters aligned with HPC's authentic, mission-driven identity
- Defined technical architecture leveraging AWS Bedrock's managed AI infrastructure for scalability and operational simplicity
MVP Development & B2B Launch (8 weeks)
- Built core conversational AI engine using Anthropic Claude with structured quote data collection workflows
- Developed Express.js backend API integrating quote calculation logic validated by HPC operations team
- Created React-based chat widget with responsive Tailwind CSS design matching HPC website aesthetics
- Deployed initial MVP focused on B2B quote automation to validate business logic and data collection flow
- Monitored early conversations closely, refining response accuracy and conversation navigation based on real partner interactions
- Iterated on conversation design to reduce friction, clarify ambiguous questions, and improve quote accuracy
B2C Expansion & Knowledge Base Integration (6 weeks)
- Extended conversational AI capabilities to handle B2C shopping assistance scenarios with product-focused dialogue
- Integrated product catalog, sustainability documentation, and customer FAQ knowledge base using AWS Bedrock Data Automation
- Implemented intelligent conversation routing distinguishing between B2B and B2C intent based on contextual signals
- Deployed unified chatbot interface serving both audiences with context-appropriate language and information depth
- Conducted user acceptance testing with HPC team and select customers to validate experience quality
- Refined natural language understanding to improve intent detection accuracy and reduce conversation friction
Analytics, Optimization & Ongoing Enhancement (Ongoing)
- Continuous conversation monitoring and quality assurance analyzing interaction logs for accuracy, brand alignment, and user satisfaction
- Regular knowledge base updates incorporating new products, policy changes, and sustainability impact data
- Iterative conversation flow improvements based on escalation patterns and user feedback
- Performance optimization ensuring sub-500ms response times during traffic peaks
- Analytics reporting providing HPC visibility into quote volume, conversion metrics, and customer insights
- Feature enhancements expanding automated capabilities based on demonstrated business value
Business impact (in progress)
This engagement is in active development. Early results show meaningful operational efficiency gains.
Early results
B2B Efficiency
Quote generation from 2-3 days to under 60 seconds24/7 global availability captures partner inquiries across all time zonesConsistent, accurate quote calculations replace manual spreadsheet processesB2C Experience
On-demand product and sustainability information during browsingProactive shopping assistance for uncertain customersAfter-hours support maintains engagement when the team is unavailableResource Reallocation
HPC team shifts from reactive inquiry handling to strategic work: partner relationships, product development, growth planningCustomer service specialists focus on high-complexity interactions where human expertise matters most
Metrics being tracked
Metrics being tracked
B2B quote request volume segmented by material type, quantity, and geographyConversion rates from automated quotes to sales follow-upOff-hours lead capture volumeProduct inquiry volume and topic distributionCart abandonment reduction where chatbot interaction occurredHuman escalation rates and patterns
Key takeaways
What's working
Discovery before code
Analyzing historical customer service transcripts and B2B emails before building showed us how people actually talk about recycling quotes and products. That shaped conversations that feel natural instead of forcing users into rigid chatbot logic.MVP-first approach
Launching B2B quote automation first validated the business logic and infrastructure under real load before expanding to B2C. Early wins built organizational confidence.Managed AI infrastructure
AWS Bedrock and Anthropic Claude eliminated months of foundational AI work. No custom model training needed. Resources focused on business logic, integration, and user experience instead.
Technical approach
The intelligent conversation routing system infers B2B vs. B2C intent from conversational context, question patterns, and linguistic cues. No explicit user self-identification required. Both audiences get specialized experiences from one unified system. This pattern works for any business serving distinct customer segments through a single interface.
Services featured in this project
Custom Software Development
Full-stack chatbot platform with React frontend, Express.js backend, and PostgreSQL databaseAI Development
Conversational AI using AWS Bedrock and Anthropic Claude for natural language understanding and context-aware dialogue
Related case studies
Real Estate / Property Technology
Reperks: Production AI Agent for Utility Billing
98% time reduction, live at agent.reperks.de
Insurance & Healthcare Technology
Signal Iduna: Digital Health Insurance Platform
300,000+ policyholders, cross-platform
Ready to build something that works?
Tell us about your project, your KPIs, and your timeline. Whether you need a conversational AI system, a dual-audience chatbot, or something else entirely, we'll help you map the path from idea to production.
Tell Us About Your Project
Describe what you're building, your timeline, your KPIs, and any technical constraints. We'll review your project and get back to you with a clear next step.Let's talk about your project