SaleMind

Inspiration

We spoke with Kapcer from Eleven Labs and several VCs about our idea. We received valuable feedback—focus on a single use case. Initially, we wanted to build a framework for Shopify (and we still plan to complete the project by the end of the month to showcase its functionality on one major platform). However, today, we decided to concentrate on something we’ve been building for years—a project where we first met. It all started with a website and now we try to build our First Voice AI Agent, built from scratch

What it does

SaleMind is a game-changer. It helps customers choose products, suggests deals, and drives sales in real time—boosting your e-commerce revenue.

Our AI Agent is a framework designed to increase sales for influencers, YouTubers, and anyone working in e-commerce. You can interact with the agent using natural voice in multiple languages to ask about products, recommendations, and actions. Who wouldn't want their favorite influencer to help them make purchases?

The agent is responsible for many tasks:

• Add products to the cart
  
• Filter products based on user preferences
  
• Complete transactions seamlessly
  
• *Using catchy phrases that are characteristic of the influencer we want to buy something from
  
• And many more - It is the Best SalesMan in town
  

This innovative solution makes online shopping more intuitive and efficient than ever before! 🚀

How we built it

• Technologies Used:
  
  • Text-to-Speech: Eleven Labs API
    
  • Speech-to-Text: Azure
    
  • Backend: Python
    
  • Integration: Multiple AI agents working together

Challenges we ran into

Everything was going perfectly until 8 PM on Saturday...

The toughest part was building everything from scratch. Every element had to be designed, optimized, and tested, requiring patience and adaptability. This was a humbling experience and a significant learning opportunity.

Accomplishments that we're proud of

• Multilingual product inquiries – Our AI Agent allows users to ask about products in multiple languages, even though the product database is in Polish.
  
• Building a user interface for the AI Agent – We developed an intuitive UI that seamlessly integrates with online stores, enhancing the shopping experience.
  
• Creating an AI Agent action list – Our agent can recommend products based on user preferences
  

What we learned

Throughout this journey, we gained significant knowledge, particularly in:

• Integrating an online store with voice – learning how to control an e-commerce platform using voice commands.
  
• Handling initial AI tasks through voice commands – a key component of our project.
  
• Optimizing WebSocket in Docker – we discovered that Docker requires a special interactive mode to ensure messages arrive immediately instead of with a delay.
  
• Approaching POC and MVP development – we used to aim for perfection from the start, which slowed down our progress. Now, we understand that rapid testing and iteration are crucial.
  

Team Members

• Kamil Woźniak – Product Owner, Data Analysis, ML
  Product management, data analysis, and machine learning.

• Daniel Durau – Voice AI Agent
  Handling and development of AI voice agents.

• Jakub Saletra – Voice AI Agent, UI/UX, Cloud
  Designing user interfaces, user experience, and much more.

• Jan Pieczywok – AI Agent Actions
  Managing and implementing AI agent actions.

What's next for SaleMind

We have a clear roadmap—by the end of the month, we aim to complete the project adding more Voice AI Agent Actions and demonstrate it on a major e-commerce platform. With the feedback we received, we can refine our product to better fit market needs.

Built With

• azure
• elevenlabs
• python
• react