LoyaltyX

A decentralized loyalty platform where business owners can register loyalty porgrams for their business, which will be operated by near-ai-assistants for seamless onboarding over the Near blockchain.

Video Demo

https://www.loom.com/share/1e5c1628515a4a23b5a18ac7c6a68c25?sid=110a83bf-e15c-4d6f-88ea-3f32e53cfee8

Inspiration

The inspiration behind this platform comes from the need to modernize and decentralize loyalty programs. Traditional reward systems are often fragmented, restrictive, and lack transparency. By leveraging the NEAR Protocol its Agentic-AI-Assistant, we empower businesses to create seamless, fair, and interoperable loyalty programs. Users can earn and redeem rewards effortlessly while ensuring security and trust through blockchain technology. Our goal is to abstract customer engagement by making loyalty programs more accessible, transparent, and rewarding for both businesses and consumers.

System Design

Getting started

Set the local LLM for the AI agent:

# Install llama.cpp
brew install llama.cpp

# Install the Hugging Face library
pip install huggingface_hub

# Login to your Hugging Face account
huggingface-cli login

# get the model from Hugging Face
huggingface-cli download NousResearch/Hermes-2-Pro-Llama-3-8B-GGUF Hermes-2-Pro-Llama-3-8B-Q4_K_M.gguf --local-dir model

# run the model with llama.cpp
llama-server -m ./model/Hermes-2-Pro-Llama-3-8B-Q4_K_M.gguf

If you want the llm to run with GPU acceleration (Optional):

Set up llama.cpp in this way:

# Clone the llama.cpp repo, and configure it for gpu acceleration
git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
mkdir build
cd build
cmake .. -DGGML_CUDA=ON
cmake --build . --config Release 

# Run the model locally at port 8080 with the llama-server
# pwd: llama.cpp/build/bin
./llama-server -m /home/<USER>/model/<MODEL-NAME> --n-gpu-layers 50

# Monitor the gpu stats
gpustat -i

Execute the near-ai-assistant backend:

# Create the suitable conda env
conda create -n ai-assistant python=3.11
conda activate ai-assistant

pip install poetry
poetry install

# Run the python backend
cd near-ai-assistant/ai
poetry run python3 src/main.py

Execute the client service

# pwd: ./LoyaltyX
pnpm i
pnpm dev

Give the llm url in file general.yaml under folder near-ai-assistant/ai/config

What it does

LoyaltyX is a decentralised loayalty program platform that allows businesses to:

• Create and manage their own loyalty programs on the blockchain
• Set custom reward thresholds and amounts
• Track customer transactions and points
• Integrate with an AI powered chat interface for seamless customer interactions

Customers can:

• Earn points across multiple businesses with a single wallet
• View their points balance and rewards
• Interact with businesses through an AI chat interface
• Redeem rewards automatically through smart contracts

Technologies used

• Hugging Face: We are running NousResearch/Hermes-2-Pro-Llama-3-8B-GGUF llm model locally on system with the help of llama.cpp with cuda-toolkit for gpu acceleration to serve as a base layer AI model for our near-ai-assitant

• Near-AI Assistant: Core of the agentic-ai

Challenges we ran into

• Handling/catching real time event emissions from the smart contracts
• Finetuning the near-ai assistant as per our platform's requirements.
• Creating a seamless UX that bridges Web2 and Web3

Benchmarks

To run quatitative bechmarks of our AI agent, run the following command:

# pwd: ./benchmarks
python3 run.py

The final metrics of latency, throughput and command_distribution benchmarks are stored in ./results folder, along with the cummulative results of the tests in ./results/final_report.csv

Contract Address

Contract address- lx9.testnet

What's next for StampX

• Multi-chain support: Expand beyond Base to include other EVM-compatible chains like Ethereum mainnet, Polygon, and Arbitrum, allowing businesses to choose their preferred network.
• Production deployment: Launch on Base mainnet with a focus on onboarding small to medium-sized businesses in the food and retail sectors.
• Enhanced AI capabilities: Implement personalized recommendation systems that analyze customer purchase history and preferences to suggest relevant products and optimize reward strategies.
• Cross-business rewards: Enable businesses to form partnerships and allow customers to earn and redeem points across multiple participating merchants.

Contributing

We welcome contributions to enhance and expand our platform! If you're passionate about Web3, decentralized applications, or event management, join us in building the future of event hosting.

Fork this repository. Create a feature branch (git checkout -b feature-name). Submit a pull request with your changes.