Technical Overview
Document Description
The following document provides a "Technical Overview" of the Cybernetic Economic Intelligence System (CEIS).
Table of Contents
List of Acronyms
List of Figures
Figure 1. Distributed Proof of Staking (DPoS) block generation within the network
Figure 2. CEIS blockchain architecture
Figure 3. Registry and resolver smart contract for CNS domain
Figure 4. CEIS DAO governance and incentive infrastructure
Figure 5. CEIS data lake architecture
Figure 6. E-commerce advisor system architecture
Figure 7. CEIS NFT sequence
Figure 8. Overall system architecture
1. Problem Statement
The CEIS is a transaction protocol that leverages AI agents and an integrated economic system for users to trade goods and services on a distributed and decentralized marketplace. AI agents are assigned to each user’s account, utilizing their funds to facilitate market transactions while generating a profit that will be returned to users in the form of passive income.
The platform leverages state-of-the-art AI/ML and blockchain use cases to facilitate automated trades of goods and services. There are also provisions for the user to avail of crypto loans based on their credit score generated from a set of predefined conditions.
From a high-level perspective, the problem can be categorized into the following:
Cybernetic Economic Intelligence System
E-commerce advisors
DAO Governance
Triptolemus and Tripto
Triptonomics Management
Monee Network
Monee Marketplace
Smart Contract Layer—Community Sourced Funds (CSF)
Credit Management—Community Sourced Funds (CSF)
2. Solution Overview
The solution consists of Blockchain, Artificial/Machine Learning, and Computer vision components as well as a cloud-hosted web application that invokes smart contracts and robo-advisory on a continuous basis. The project begins with the launch of the platform’s native token, Tripto, a custom token built on top of the proof-of-stake blockchain network.
Product listings will consist of NFTs, preferably of the ERC-721 Standard. To facilitate some of the above-mentioned functionalities, we will be developing computer vision algorithms for image matching between two listed products, as well as a recommendation engine to match the right buyer with the right seller.
We will be hosting a platform token oracle to exchange fiat with Tripto, in order to streamline transactions for novice crypto users and external vendors who may have hesitancy incorporating crypto into their balance sheets. Still, Tripto will be the sole currency used to facilitate commercial trades across the platform.
We are not considering a KYC module to ensure anonymity and true decentralization, but there will be provisions to track users and avoid misuse. There will also be a credit score module that generates a unique signature-based credit score for all users.
The DAO-based investment scheme offers investors a lucrative opportunity to gain revenue from the order matching module, which is fulfilled through each account’s AI agent. There will be an admin master wallet and a DAO wallet that will act as the primary store of funds on the platform.
2.1. Blockchain Architecture
The blockchain network uses the Delegated Proof of Stake (DPoS) consensus protocol as the platform’s staking feature. Block generation is performed through delegates that are randomly selected by the system. Stakeholders are composed of network users whose AI agents allocate a portion of their investment funds into various staking CSFs. Stakeholders are thus comprised of CSF holders, while delegates from those funds are randomly selected by the system to validate and verify various operational transactions (i.e. product order confirmations, warehouse receipt of shipments, yield distribution, etc.). Rewards for block generation are distributed according to the stake each investor has contributed to the fund. Stakeholders hold no more or less decision power based on the amount they have staked in the fund.
The system’s security is critical to the blockchain network, as well as its native token. This is why it uses elliptic curve cryptography and cryptographic hashing. The cryptographically secure key pairs are generated using elliptic curve cryptography and consist of private and public keys. During block creation, delegates take a fixed number of transactions from the transaction pool and confirm/validate key transactional activities. The confirmed transactions will be added as payload to the block and signed by the delegate. The process of signing a block header is the same as the process of signing a transaction. The signing is done using a delegate secret key, and the header is hashed using SHA-256.
Figure 1. Distributed Proof of Staking (DPoS) block generation within the network
Nodes in the network use Remote Procedure Calls (RPC) events to communicate block and transaction propagation. RPC events also transmit JSON objects with additional fields for processing the transmission. To effectively transmit the JSON object, the network uses web socket. For node communication with other networks, the system header will be used to identify them (nodes) and acquire basic information.
Figure 2.CEIS blockchain architecture
Blocks are generated in a decentralized way, being sent back to other nodes for further propagation. The block creation will select a set of nodes in the network and broadcast the transaction. Those that receive the block will append it to their ledger and further broadcast it to another few random nodes. All transactions from the node will be first collected in a transaction pool, which acts as a memory bank until it is added to the block. Transactions are moved from one node to all other nodes in order to be included in blocks. The broadcast queue for transactions functions by drawing up to a fixed number of transactions from the transactions pool and performing a validation process of those transactions.
3. Cybernetic Economic Intelligence System
The CEIS dynamically matches acquired information from selected actions relative to a computational issue that defines the essential purpose of the system or machine. In our use case, this module will host data ingestion and processing pipelines in order to generate actionable insights that will be used to manage the DAO with the help of AI agents.
The system’s architecture is designed to construct, distribute, and maintain the facilities of user accounts. Each account’s respective AI agent will be able to make automated trades on the platform based on their contribution to fractionalized NFTs, called Community Sourced Funds. This enables AI agents to automate commercial processes and receive perpetual yield payments from the profits of each marketplace sale.
The module is thus divided into the following:
Admin Section
User Section
In the Admin Section, there will be a provision for the platform administrators to view and modify the current Tokenomics set. They will be able to execute pertinent decisions, such as adjusting the fees and commission % for users, manufacturers, distributors, etc.
Tokenomics features, such as inflation control for platform tokens, will be pre-defined and deployed on the token contract itself and will be unchangeable after the token contract has been launched.
The CEIS platform will have a data processing layer that consists of an ETL section (‘Extract, Transform, and Load’). To facilitate data storage and analysis for the purpose of extracting insights, we will be using Azure infrastructure provisions.
3.1. Domain Management for User Smart e-Businesses
Crypto Name Service is a set of smart contracts on the blockchain that govern how domains are created and used. Although its purpose is similar to that of a traditional DNS system, CNS has architectural differences that change the interaction model significantly. For example, CNS domains are owned irrevocably. They do not need to be renewed and cannot be reclaimed. Once minted, members have complete control of their domains.
Every CNS domain will be issued as an ERC-721 token. Building on this standard makes it easier for developers to integrate with the platform and allows members to manage their domain ownership with compatible wallets, exchanges, or marketplaces.
The two central components of CNS are its Registry and Resolver smart contracts. Registry is a map (or dictionary) from domain names to an owner address and a Resolver address. Resolver is a map from domain names to the records associated with that domain (cryptocurrency addresses, etc.).
There will only be one Registry smart contract deployed in the blockchain network, however, there are many versions of Resolver smart contracts. In theory, every domain could use a different Resolver contract, but in practice, most domains are managed by the same Resolver smart contract instance.
Figure 3. Registry and resolver smart contract for CNS domain
3.2. Info
Updates to our Resolver smart contract are incremental and non-breaking. All Resolver smart contracts must adhere to our IResolver interface. This interface defines the basic operational functionalities and guarantees compatibility between different implementations.
Each ERC-721 token is identified by a unique number, its tokenID. To make domains identifiable, we’ll use a process called Namehashing.
3.3. The CEIS: A Decentralized Autonomous Organization
The CEIS is an open-source Decentralized Autonomous Organization (DAO) created in a Proof of Stake custom blockchain for the governance and incentivization of CEIS ecosystem participation.
Figure 4. CEIS DAO governance and incentive infrastructure
3.4. Tripto
Tripto enables community ownership and active stewardship of the CEIS ecosystem.
Total supply
Token distribution
Token vesting
3.5. Tripto Holders
Blockchain Address in possession of Tripto.
Network reward distribution
Transfer between accounts
Token/fiat exchange via the platform’s token oracle
3.6. Proposals
Proposals are the means by which Tripto holders within the CEIS network are able to make changes to the ecosystem.
Proposals are classified based on the mode of execution.
Executable proposal
Social proposal
Proposals are classified based on the type of voting. Each vote is counted as 1 vote per voter.
Proposal quorum
Proposal queue
Proposer minimum requirements.
Proposer minimum votes.
Proposer minimum delegate votes.
Proposals are classified based on their impact type.
Technical
Funding
AI engine score calculations
Policy change, etc.
Storage of proposals on the blockchain
Checksum of the proposal stored in the blockchain
IPFS hash of the proposal stored in the blockchain
3.7. Voting
Voting is the process by which Tripto holders vote for or against an active proposal.
Types of votes
Direct vote
Delegate vote
Delegate by signature
Voting delay
Voting period
4. Data Lake Architecture
This represents how the system’s Data Lake will be implemented on Azure. The three main features of Azure we will be utilizing for this use case are:
Azure Data Service,
Azure Query service, and
Azure Data storage.
Azure SSIS is a serverless data integration service that makes it easy to discover, prepare, and combine data for analytics, machine learning, and application development. It provides all the capabilities needed for data integration so that data can be analyzed.
Figure 5. CEIS data lake architecture
Data Lake Analytics is an interactive query service that enables data analysis in Azure Blob Storage using standard SQL. Data Lake Analytics is serverless, so there’s no infrastructure to manage.
Azure Storage account, or Azure Blob Storage, is a service offered by Azure Web Services that provides object storage through a web service interface. Azure Blob Storage uses the same scalable storage infrastructure that Azure.com uses to run its global e-commerce network.
With the above three components, we will be able to collect, process, and generate insights from user data, which will allow us to serve customers better by understanding user actions.
The User Module will include functionalities for wallet connection, allowing users to connect wallets such as MetaMask to transact on the network. When users log in by connecting their wallets, they will have the option to purchase DAO tokens using native cryptocurrency.
5. E-Commerce Advisor System Architecture
AI agents are user-owned autonomous intelligent agents that understand marketplace scenarios on the platform. They will have access to the funds collected from users, allocating them as capital for investment in CSFs.
Agents will be powered by python scripts. An intent detection module identifies the correct action to take based on the input received from the user through the CEIS module. The Nginx server is being implemented to improve the scalability of the platform. Entity recognition is useful for processing documents from manufacturers to whom AI agents will be placing orders. This requires a document processing engine to understand the invoices generated from manufacturers and retailers.
A small talk engine will be introduced to make the agent behave as a person would. Users will be able to ask out-of-context questions and the agent will be able to respond with a meaningful response.
Figure 6. E-commerce advisor’s system architecture
Action Engine is where the AI model performs regression or classification based on inputted data, outputting a decision response. It receives inputs from the information extracted from the market and analyzes historical market values. The action engine model has a cluster of models whose individual responses infiltrate the final decisions. This will then be processed by the response selector, along with the time series forecasting of the product whereby the AI agent can decide the best values for decision-making (i.e. based on historical trends, when is the best time to buy/sell a product and for what price). AI agents are uniquely designed to communicate with other agents with similar motivations that are characterized by the user’s data/investment profile. Their inter-communication is for the purposes of obtaining a better understanding of the market and is ongoing until a final decision has been made and executed within the action engine.
We may also extract information from global scenarios. Our AI model will observe international news in order to forecast major fluctuations, such as supply shortages, using advanced NLP and entity recognitions to collect relevant real-time information. As the model is fed, it will compile a number of strategic outputs that far exceed any derived from the human mind, with greater accuracy. This model will analyze and assign specific weightings for different product categories in the marketplace, assessing market changes and how various sectors correlate with one another. AI agents will be generally designed for reinforcement learning, constantly learning from their historical decisions in order to optimize future accuracy.
6. DAO Management
On the Moneetize platform, every user will have a self-managed wallet that will be accessible to their AI agent. A user will be able to add funds to this wallet and receive the platform’s native currency as investment capital.
There are 2 sections to DAO management:
the platform accessible to the admin, and
the platform accessible to users.
In the DAO section, users will be able to connect to their wallets and raise or vote on proposals. The DAO administration will be able to view the ongoing list of proposals, as well as their voting outcome in order to deliberate their implementation.
7. Triptolemus and Tripto
Triptolemus is the name of the central logic module of the CEIS. It hosts a mathematical framework that functions as both the economic intelligence and policy moderator for each user account, as well as the network economy as a whole. For the user, it functions as a real-time reference for market dynamics, signaling points of disequilibrium in supply, demand, or pricing. Similarly, it’s also designed to maintain a symmetrical distribution of wealth across the entire network while simultaneously increasing the productivity, wealth, and income of the entire network[1].
Tripto will be deployed on the Polygon Network to capitalize on high transaction speed and low gas fees, ideal for our use case.
8. Triptonomics Management
There will be an admin-specific feature with provisions to tweak parameters around network transactions. There will also be an analytics feature, where information such as the current price, the total number of tokens in circulation, the biggest token holders, information about the DAO holders, and the number of active proposals can be viewed.
This section of the solution will be a generic web application powered by React JS for the front end and NodeJS for the back end, with PostgreSQL as the database.
9. Monee Network and Marketplace
The Monee network behaves as a collaboration network between Triptolemus and user DAOs. It provides AI agents that manage user smart e-businesses the ability to talk to each other. This will be facilitated by an inter-process communication protocol.
Monee Marketplace is a unique implementation of product listings and services. Marketplace products are listed on-chain via NFT minting (standard ERC-721).
Figure 7. CEIS NFT sequence
The above sequence diagram explains how manufacturer product listings undergo the lazy minting process.
Another feature we will be incorporating here is a computer vision-based AI module that will find the similarity between two listings and group them together. This will allow us to perform product recommendations, which will then be used by AI agents to facilitate trades.
10. Smart Contracts and Credit Community Sourced Funds
Community Sourced Funds are a financial empowerment mechanism for both types of users on the network (consumers and manufacturers).
For consumers, they enable the diversification of investment opportunities, diluting risk while simultaneously increasing returns.
For manufacturers, CSFs expand their market opportunities, attracting clients from all over the world to syndicate investment in their products.
The goal of this layer is to facilitate trust between multiple parties in the system. There will be a provision for users to add the product listing pricing and other conditions, such as commission fee, royalty, etc. All of which will be enforced by the smart contract.
This section also has crypto lending, enabling producers to request funds for working capital requirements. This will be a collateral-based system where the user has to enter the amount of crypto needed and the platform will process it. Based on the customer’s credit history, it will calculate the collateral needed for issuing the credit.
The payouts will be in Tripto, which users can liquidate through the token oracle or third-party exchanges.
11. Overall System Architecture
Figure 8. Overall system architecture
11.1. Architecture Description
The application architecture consists of the following components:
Azure DNS
Azure CDN
Azure Blob Storage
Application Load Balancer
Azure Shield (DDos Protection Service)
Azure Web Application Firewall
Azure Service Bus
VPC
Availability Zones
Public Subnet
Private Subnet
Virtual Network NAT
Virtual Machines
Azure SQL
Redis Cache
Client Module
Blockchain Module
3rd Party Modules
The application will be available through a web interface and communicated to the backend through REST API calls. The application will be hosted on Azure cloud and incorporates the above-mentioned features to ensure scalability, security, and minimum downtime.
Azure DNS handles the DNS. Azure CDN is a content delivery network operated by Azure Web Services. We will be providing static website assets through CloudFront, coupled with a Blob Storage that stores static assets.
Since we’re focusing on maximum uptime, it’s best practice to make the application available on multiple Azure regions. That’s why we’ll introduce a load balancer to effectively manage and route incoming traffic. Azure Shield is a managed Distributed Denial of Service (DDoS) protection service that will safeguard the application. Azure WAF is a web application firewall that helps protect the web application against common web exploits and AI agents that may affect availability, compromise security, or consume excessive resources. Azure WAF provides us control over how traffic reaches the application
Azure Service Bus is a notification service that provides a low-cost infrastructure for the mass delivery of messages and emails to users. A Virtual Private Cloud (VPC), an on-demand configurable pool of shared resources allocated within a public cloud environment, provides a certain level of isolation between different resources. This introduces a modular security architecture that helps contain breaches.
Availability Zones are distinct locations within an Azure Region that are engineered to be isolated from failures in other Availability Zones. They provide inexpensive, low-latency network connectivity to other Availability Zones in the same Azure Region. Each region is completely independent.
If a subnet is associated with a routing table that has a route to an internet gateway, it’s known as a public subnet. If a subnet is associated with a routing table that does not have a route to an internet gateway, it’s known as a private subnet. We’re incorporating subnets to maximize security within the architecture.
Virtual Machines are a part of Azure’s cloud-computing platform, Azure Web Services, which enables users to rent virtual computers on which they can run computer applications.
Azure Relational Database Service is a distributed relational database service. It is a web service running ‘in the cloud’ that helps simplify the setup, operation, and scaling of a relational database for use in the application. We will be using a managed PostgreSQL instance powered by RDS. A Redis cache will be introduced to cache database responses and API responses in a key-value format, improving the scalability of the platform.
Third-party modules include CoinGecko to fetch real-time token price information and Sendgrid for sending emails. This section is kept modular to accommodate future 3rd party integrations, such as KYC, Payment processor, etc. The blockchain module consists of the smart contracts that will be written on Solidity and deployed to an EVM-compatible public chain, such as Ethereum or Polygon.
To learn more about Triptolemus’ economic intelligence; see our ‘Triptonomics’ white paper. ↑
Last updated
