​
Architecture Overview

The AGIXBT Framework consists of several key components, each serving a distinct purpose in the overall functionality of the agent.

​
1. Data Ingestion Layer

  • Twitter Scraping API: Implements a scraping mechanism to gather tweets from top ecosystem accounts.

    • Scraping Mechanism: Utilizes both API-based and web-scraping techniques to fetch tweets, ensuring compliance with Twitter’s platform terms of service.

    • Data Filtering: Filters out irrelevant noise and identifies high-value information using Natural Language Processing (NLP) techniques.

  • Data Ingestion Language Models: Processes raw scraped data through multiple Language Models (LLMs) to clean, categorize, and enhance the information.

    • LLM Selection: Chooses specialized models for Open Source Intelligence (OSINT) tasks to ensure accuracy and relevance.

    • Data Contextualization: Enriches data by integrating on-chain activity and other relevant sources.

​
2. AI Processing Layer

  • Context Windows: Defines adjustable windows for maintaining conversation context, enabling the agent to understand and respond to user interactions coherently.

  • Messaging Language Models: Handles the generation of tweets and replies based on the processed data and interaction triggers.

    • Response Generation: Uses prompt engineering to craft contextually relevant and engaging posts.

    • Trigger Mechanisms: Activates only when tagged or directly addressed, minimizing unnecessary interactions.

​
3. On-Chain Activity Pipeline

  • On-Chain Data Integration: Pulls relevant blockchain data to supplement tweet content, ensuring up-to-date and accurate information.

    • API Selection: Chooses reliable blockchain data providers for robust integration.

    • Data Query: Executes specific queries to retrieve on-chain information pertinent to the agent’s ecosystem.

​
4. User Interaction Layer

  • Twitter Posting Functionality: Posts curated information on Twitter based on predefined characteristics and scraped data.

    • Content Strategy: Uses character cards to tailor the tone and style of posts.

    • Scheduling: Implements a scheduling system for optimal engagement timing.

  • Reply Functionality: Engages with users when tagged, responding to questions and maintaining a coherent dialogue.

    • Trigger Mechanisms: Only activates when mentioned, ensuring focused interactions.

    • Contextual Responses: Uses stored context to provide accurate and helpful replies.