Inspiration
What it does
How we built it
Inspiration
In many parts of Africa, small businesses don’t rely on websites — they sell through chat.
From WhatsApp to Instagram DMs, customer conversations are where business happens. But this creates a major problem: business owners can’t respond to every customer instantly, follow up consistently, or manage multiple conversations at scale. This leads to missed sales, poor customer experience, and lost revenue.
I wanted to solve this by asking a simple question:
What if every small business had an AI sales agent that could chat, recommend, and close deals automatically?
That idea became ConversaShop AI.
What it does
ConversaShop AI is a conversational commerce platform that allows businesses to sell products entirely through AI-powered chat.
It acts like a virtual salesperson that:
- Responds to customer inquiries in real-time
- Recommends products based on user preferences
- Handles negotiations and budget constraints
- Converts conversations into actual orders
- Simulates mobile money payments
- Tracks inventory and updates stock automatically
The platform also includes a business dashboard where owners can:
- View orders and sales
- Manage products and inventory
- Analyze customer behavior
- Receive AI-generated business insights
A unique feature is the Live Simulation Mode, which demonstrates multiple customers interacting with the AI simultaneously — showing how the system scales in real-world scenarios.
How I built it
This project was built using MeDo as the core development engine.
I leveraged MeDo to:
- Generate a full-stack application (frontend + backend)
- Design a multi-agent AI system, including:
- Sales Agent (product recommendations & persuasion)
- Order Agent (checkout & order processing)
- Support Agent (FAQs & issue handling)
- Inventory Agent (stock validation)
- Implement a multi-turn conversation engine with:
- Intent detection (buy, browse, negotiate, complain)
- Context memory across conversations
- Dynamic response generation
- Create a modern, responsive UI using Tailwind CSS
- Build a simulated chat environment inspired by WhatsApp
I also extended functionality with:
- Simulated mobile money payments
- Real-time analytics dashboard
- AI-powered insights and recommendations
Challenges I ran into
One of the biggest challenges was designing a conversation system that feels natural and human-like while still being structured enough to handle real business logic.
Another challenge was coordinating multiple AI agents to work together seamlessly. Ensuring that the system could route conversations correctly between sales, support, and order handling required careful prompt design and state management.
Balancing technical depth with a smooth and intuitive user experience was also a key challenge.
Accomplishments that I'm proud of
- Successfully building a multi-agent conversational system
- Creating a realistic AI sales experience that can actually convert conversations into orders
- Designing a clean, mobile-first UI that feels familiar to users
- Implementing a live simulation feature that showcases scalability
- Demonstrating real-world impact for small businesses
What I learned
- How to structure advanced multi-turn AI conversations
- How to design modular AI systems with multiple agents
- The importance of UX in AI-powered applications
- How to translate real-world problems into scalable software solutions
What's next for ConversaShop AI
- Integrating real WhatsApp Business API
- Adding real mobile money (MTN, Airtel) payments
- Expanding to multilingual support
- Deploying for real small businesses across Africa
The long-term vision is to build a full conversational commerce ecosystem where businesses can operate entirely through chat.
Built With
- ai
- css
- javascript
- llm
- medo
- node.js
- rest
- sqlite
- tailwind
Log in or sign up for Devpost to join the conversation.