> For the complete documentation index, see [llms.txt](https://whitelistzone.gitbook.io/whitepaper/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://whitelistzone.gitbook.io/whitepaper/vision-and-solution/solution-architecture-technical-deep-dive.md). # Solution Architecture (Technical Deep Dive) **WhiteList Zone** is a hybrid swarm of AI agents that combines Web3 and ML: **Data Layer**\ πŸ”Ή **AI Scrapers Γ—10**\ – Scrape from 10+ sources: Discord, X (Twitter), OTC platforms, Alphabot, Subber, Atlas3, Premint.\ – Update speed: every 15 minutes (6Γ— faster than manual methods). πŸ”Ή **Swarm Agents**\ – 12 autonomous models:\ β†’ NLP classifiers for finding WL in 10 languages.\ β†’ GNN (Graph Neural Networks) analysis to detect connections between projects and scam schemes.\ – RAG (Retrieval-Augmented Generation): real-time data updates via distributed GPUs (ElizaOS). **AI Layer**\ πŸ”Ή **Scam Detection Engine**\ – Analyzes 50+ parameters: from GitHub activity to tokenomics.\ – Outcome: filters out 80% of risks by identifying scam patterns. πŸ”Ή **PNL Predictor**\ – Forecasts ROI using an LSTM model with a target accuracy of 80% (trained on historical public IDO data from 2021–2023). πŸ”Ή **Dynamic Pricing Core**\ – An algorithm that adjusts the NFT box price daily based on:\ β†’ Project’s social activity (Sentiment Analysis).\ β†’ OTC demand (Whales Market, Exsaverse, Telegram OTC). **Execution Layer**\ πŸ”Ή **NFT Boxes (ERC-1155)**\ – Execution guaranteed by smart contracts (integrated with Chainlink Oracle).\ – Example: WL of project X sold for 0.5 ETH β†’ 3 secondary market deals within 24 hours (+20% in price). **Visualization:**\ An architecture diagram (3 blocks: Data β†’ AI β†’ Execution) with arrows:\ – Data Sources β†’ AI Scrapers β†’ Swarm Agents β†’ PNL Predictor β†’ NFT Boxes β†’ Marketplace.\ – Captions: β€œEvery WL goes through 4 stages of AI verification.”\ β€œEffect for projects (modeling):\ – CAC: reduced by 60% ($200 β†’ $80)\ – Conversion: 90% of slots go to real users.” **Case Study in the corner of the slide:**\ – β€œHow it worked for project X: CAC reduced by 60% β†’ $200β†’$80, 90% of slots purchased by real users.” **Tech stack (in small font at the bottom):**\ ElizaOS, Chainlink, GNN, ERC-1155, LSTM.