Digital Transformation: Design and Implementation of a Blockchain Platform for Decentralized and Transparent Property Asset Transfer Using NFTs

Abstract

In many jurisdictions, property registration and transfers remain constrained by inefficient, paper-based processes that depend on multiple intermediaries and bureaucratic approvals. This paper proposes a decentralized, blockchain-based property platform designed to streamline these processes using Non-Fungible Tokens (NFTs) and artificial intelligence (AI) agents to modernize public-sector asset management. The work addresses the persistent inefficiencies of paper-based property registration and ownership transfer by embedding legal and administrative logic within smart contracts and automating compliance through an intelligent conversational interface. The system was implemented using Ethereum-based ERC-721 standards, React for the user interface, and Langfuse-powered AI integration for guided user interaction. The pilot implementation presents secure, transparent, and auditable property-transfer transactions executed entirely on-chain, while hybrid IPFS-based storage and decentralized identifiers preserve privacy and legal validity. Comparative analysis against existing national initiatives indicates that the proposed architecture delivers decentralization, citizen control, and interoperability without compromising regulatory requirements. The system reduces bureaucratic overhead, simplifies transaction workflows, and lowers user error risk, thereby strengthening accountability and public trust. Overall, the paper outlines a viable foundation for legally aligned, AI-assisted digital property registries and offers a policy-oriented roadmap for integrating blockchain-enabled systems into public-sector governance infrastructures.

1. Introduction

The digital transformation of the public sector is a strategic necessity, especially in the context of an increasingly interconnected and technology-dependent society. Digital transformation ensures government transparency by allowing citizens to better understand and monitor activities in their relationship with public administration [1]. In this process, emerging technologies are profoundly modernizing public administration. Every country faces different challenges, from culture to infrastructure, so blockchain technology offers a unique and personalized way to modernize the state.

Transferring cars, real estate, or valuables from one owner to another is fraught with complex bureaucratic regulations. In most countries, selling or donating these types of assets requires extensive paperwork, civil servants, and lawyers to verify these documents. In terms of public management, the use of blockchain changes the way institutional resources, services and processes are managed. Blockchain transparency increases the accountability of public officials and gives citizens greater control over how public resources are managed.

Cutting-edge technologies such as blockchain and AI agents are changing the way governments manage public records. In a decentralized ledger system, blockchain provides a secure, transparent, and immutable framework for storing and validating data [2]. This technology eliminates the need for a central point of control, reducing the risks of corruption, fraud, and data loss. This ensures real-time access to verified information. Intelligent agents can automate traditional bureaucratic processes. Through rapid and accurate data analysis, AI can identify errors, omissions, or potential fraud, reducing the time required for manual checks and human intervention [3]. Over time, agents become increasingly efficient through continuous learning to manage large volumes of government data beyond human potential.

Integrating blockchain with AI agents allows for the creation of an administrative ecosystem with a high degree of autonomy and transparency. This contributes to a significant reduction in administrative formalities and operational costs, both for citizens and for state institutions. The use of blockchain-based tokenization helps overcome the problems currently facing the global financial system [4]. Fractional property purchases will be facilitated by tokenizing assets, these assets can be sent and exchanged with the highest level of security using a blockchain wallet [5].

This paper proposes a novel approach for developing a decentralized property registry platform that streamlines property transfer processes. The novelty of our approach lies in the explicit coupling of blockchain-based via non-fungible tokens (NFTs) driven process automation, the proactive accommodation of regulatory constraints, and a modular microservices architecture for integration into public services. In this regard, the Framework outlines the stages of establishing and operating a decentralized registry, while emphasizing continuous stakeholder engagement and legal compliance assessment. The Method demonstrates how practitioners can conduct these phases by combining smart contract execution with intelligent agent assistance, enabling iterative refinement and guiding the platform toward full operational maturity. The approach undergoes evaluation through a six-month controlled pilot study and interdisciplinary expert review, combining technical validation with legal and policy-oriented analysis to assess its real-world feasibility. During the pilot phase, preliminary qualitative feedback was obtained from 20 domain participants, whose perspectives helped refine the system’s workflow and regulatory alignment. From a broader perspective, these evaluations indicate that the platform has potential to reduce bureaucratic complexity in property transactions, while also highlighting the need for continued refinement of certain legal and technical concepts to reach full institutional consensus. Additionally, qualitative insights corroborate the practical relevance of the system and offer recommendations for improving interface clarity and documentation, further strengthening the platform’s suitability for future public-sector integration.

Traditional property-registry processes remain slow, paper-dependent, and institutionally fragmented, requiring multiple intermediaries and manual verification steps that introduce delays, errors, and limited transparency. Existing blockchain-based pilots are typically implemented on permissioned ledgers, rely on centralized override mechanisms, and they rarely integrate AI-supported user interaction. As a result, a clear gap persists between technical feasibility and citizen-oriented, verifiable, and autonomous property-transfer systems.

To address this gap, the paper introduces a decentralized property-transfer framework in which NFT-based digital twins interact directly with on-chain execution, reducing reliance on intermediaries and enhancing auditability. The main contributions of this paper are:

• Deployment on a public, permissionless blockchain (Ethereum) rather than a centralized solution used by public administration or a government ledger but permissioned. This ensure transparency, immutability, and citizen-controlled state transitions.
• Encoding all legally relevant transfer logic directly into smart contracts, without centralized override paths, enabling trustless and auditable asset management.
• Representing real-world property as NFTs functioning as digital twins, maintained in synchronization with a regulated registration process.
• Integrating a Langfuse-orchestrated agentic AI assistant, capable of guiding users through complex legal steps, validating actions, and potentially reducing operational errors integrates a Langfuse-based AI assistant that guides users through legally structured workflows and reduces operational errors.

The remainder of this paper is structured as follows. Section 2 reviews the existing literature on blockchain-based registries, NFTs, public-sector digital transformation, and related international pilot projects. Section 3 presents the system architecture, smart-contract design, and implementation details. Section 4 reports the experimental results, including performance benchmarks, gas-usage evaluation, and the user-study assessing the AI assistant. Section 5 provides a discussion of the implications, limitations, and comparative positioning of the proposed framework. Finally, Section 6 concludes the paper and outlines directions for future research.

2. Related Work

2.1. Blockchain Technology

Blockchain technology provides a decentralized, tamper-resistant infrastructure that enables transparent and verifiable state changes without reliance on a central authority, making it particularly suitable for managing high-value assets and institutional workflows such as property registration. Blockchain technology is an advanced form of Distributed Ledger Technology (DLT), in which digital data is stored in interconnected blocks [6]. Each block is cryptographically linked to the previous one, forming a sequential structure similar to a chain, and new blocks can only be added at the end of it. Ethereum is the most widely used blockchain platform for smart contracts [7], providing a robust and flexible platform for developing decentralized applications and automating transactions without intermediaries. Distribute nodes of the blockchain network process, encode and organize the received transaction data into structured blocks. Each block is then timestamped and securely attached to the main chain, thus contributing to the extension of the longest and most valid blockchain. The blockchain system uses a hashing technique and a specific data structure, such as a binary hash tree or a Merkle tree, to arrange the data into blocks. [8]. A block in the blockchain once validated and added cannot be modified, any modification would trigger a chain effect and affect all subsequent blocks making the blockchain tamper-proof [9]. Thus, blockchain technology provides a secure framework for storing and transmitting sensitive information, such as identity and property information.

The rapid development of platforms like Ethereum encourages the rapid advancement of blockchain technology [10]. Blockchain has already gained popularity in the financial sector because it allows for the safe, reliable, and verifiable conduct of financial transactions, as demonstrated by the emergence of cryptocurrencies [11]. An important aspect is its anti-double-spending feature, which guarantees that a person transfers an asset in the form of a transaction, is a component of blockchain technology in the financial sector [12]. Another aspect is that a blockchain manages and stores data, which makes it difficult or impossible to modify, hack, or defraud the network [13]. Blockchain technology can give existing banking systems a competitive edge in terms of transaction security at a reduced cost [14].

2.2. Blockchain-Based Digital Transformation of the Public Sector

Blockchain adoption in public administration has grown significantly as governments explore digital technologies to modernize registries, improve transparency, and reduce administrative burden. Blockchain technology in public administration automates processes through smart contracts. Automation reduces human errors, accelerates service delivery and contributes to the overall efficiency of the public system. Thus, blockchain technology can improve data sharing between agencies, eliminate redundant verification steps, and support secure and tamper-resistant public records [6]. These capabilities make blockchain a strong candidate for modernizing public registries, auditing processes, and public-service delivery through immutable and verifiable data structures.

Beyond the technical attributes of immutability and distributed trust, blockchain also introduces new governance models that shift value creation and decision-making away from hierarchical intermediaries toward more decentralized, community-driven configurations. Peer-based governance systems demonstrate how blockchain facilitates transparent coordination and bottom-up value distribution mechanisms, offering insights relevant to next-generation digital public infrastructures [15]. Blockchain technology in government must be understood through a socio-technical perspective. Thus, conceptual frameworks highlight gaps in readiness, interoperability, regulatory alignment, and user acceptance, which governments must address before achieving large-scale implementation [16]. Interest in blockchain-based public services has grown, but persistent challenges have been identified, including scalability constraints, integration with legacy infrastructures, and the need for standardized governance protocols [17]. The successful implementation of blockchain frameworks requires clear institutional roles, transparent accountability structures, and mechanisms to balance decentralization with public sector oversight [18].

2.3. NFTs and Tokenization Models for Asset Ownership

NFTs are unique digital assets that attest to ownership of a digital or physical good that cannot be replicated or replaced. NFTs were the first application of blockchain technology to gain widespread recognition [19]. NFTs are created on a blockchain network through a mechanism known as minting, which creates a unique token using a smart contract [20]. The intrinsic features of NFTs set them apart from traditional coins [21].

Traditionally, NFTs have been associated with digital art, virtual collections or games, but their potential extends far beyond these areas. Through blockchain trade contracts, buyers of digital assets can resell NFTs, guaranteeing dependable trading and liquidity [22]. Proving ownership in traditional systems was an unsolved problem until the advent of blockchain-based NFTs [23]. NFT market activity exhibits distinct relationships with major blockchain ecosystems such as Bitcoin and Ethereum, indicating that NFT ownership models are influenced by broader network effects, investor behavior, and platform-specific characteristics [24]. Any change in ownership such as sale, inheritance, donation can be automatically recorded on the blockchain by transferring the token, providing a clear, up-to-date and tamper-resistant record. Unlike traditional authentication methods, NFT-based identity would be impossible to steal or duplicate, thanks to the cryptographic record and immutability of the blockchain. NFTs can play a key role in this transition towards a transition to digital public administration by transforming the citizen-state interaction.

Beyond market structures, NFTs have also been directly connected to digital land and property management. Virtual real estate markets using NFTs reveal how scarcity, market visibility, and platform reputation shape the pricing of NFT-based land assets, reinforcing the idea that tokenization can be used to represent and transact forms of property value in both virtual and real environments [25].

NFTs are a convergence point between cryptographic scarcity, digital identity, and decentralized market infrastructures, highlighting their potential to formalize ownership in digital ecosystems [26]. Based on these insights, NFTs are increasingly viewed not only as tradable collectibles, but also as transferable rights associated with digital assets, legal documents, and real-world property. Legal scholars emphasize that linking digital assets to NFT identifiers necessitates robust frameworks for authenticity verification, applicability, and metadata integrity, particularly when working with regulated property or documents [27]. NFTs have been explored as a mechanism for decentralized content and asset management. New architectural models combine NFTs with decentralized storage systems such as IPFS to ensure that digital content associated with an NFT remains tamper-resistant, independently accessible, and cryptographically verifiable [28]. Thus, NFT-based property can be integrated with secure off-chain data storage, enabling scalable and regulatory-compliant ledgers for digital assets.

2.4. Blockchain-Based Land and Property Registries

Implementing and integrating a decentralized registry system in a real government context requires addressing complex legal and political challenges. One of the most significant issues is the legal recognition of on-chain transactions and NFT-based ownership. In many jurisdictions, land and vehicle registries are governed by statutes that specify paper documentation, notarization, and centralized, government-maintained ledgers as requirements for valid transfer. The current implementation, while technically robust, would require legislative reform to be recognized as legally binding and for the pilot phase the NFTs should be seen as digital twins.

In the context of the European Union, recent digital identity and e-government directives (such as eIDAS) open pathways for integrating cryptographic signatures and digital records into official processes, but specific legal frameworks for blockchain-based property transfers remain under development. Romania, for example, is actively digitizing its governmental services but has yet to introduce legislation recognizing smart contracts or NFT-based asset records as legally binding for real property. The implementation presented here could serve as a testbed for informing such reforms, demonstrating the technical feasibility, security, and auditability required by lawmakers.

The policy environment in other countries provides valuable context. Estonia, often cited as the global leader in e-government, has built a digital land registry system leveraging blockchain-inspired technologies, but the infrastructure remains permissioned and ultimately centrally managed by the government [29]. While the system delivers dramatic efficiency gains, it does not eliminate the need for institutional trust and manual verification. The United Kingdom’s Land Registry Digital Street pilot explored blockchain for improving transparency and auditability but stopped short of enabling direct, trustless transfers of title between citizens, citing regulatory and user experience challenges [30].

The Georgia pilot project provides additional information by implementing a blockchain-based land registry using the Bitfury blockchain, primarily to provide audit trails for property records. The goal of the implementation was to reduce fraud, but the system is a hybrid that links public blockchain entries with government-run databases without being a decentralized registry [31]. Dubai has similarly announced blockchain-based land registry initiatives, but these too rely on permissioned ledgers controlled by the Dubai Land Department, which preserves a centralized trust model [32].

The solution described in this article departs fundamentally from these precedents by deploying all critical data and logic on a public, permissionless blockchain, and by using NFTs to directly model property rights without the need for a trusted intermediary. This architectural difference has profound policy implications removing single points of failure and improving institutional trust.

Effective land administration systems depend on the integrity, authenticity, and transparency of property records. Distributed ledgers could enhance trust in land registration systems by providing tamper-proof audit trails and verifiable transaction histories, while also highlighting that organizational, legal, and governance challenges must be addressed before achieving full adoption [33].

A case study in the context of real estate transactions in Serbia demonstrates how permissioned blockchain architectures can secure land transactions, reduce fraud, and enhance registry reliability by ensuring that every update is cryptographically verifiable and transparently recorded [34]. The study highlights that blockchain is particularly effective in environments where trust between actors is limited, as it provides decentralized validation without relying on a single institutional authority.

A blockchain-based framework proposed for rural and agricultural property markets—where ownership disputes, incomplete documentation, and informal transactions are common—demonstrates how decentralized registries can enhance transparency, reduce administrative delays, and establish trustworthy records for communities with limited institutional infrastructure [35].

The Land Administration-as-a-Service (LAaaS) model proposes cloud-based, integrated, and interoperable land-information systems capable of supporting blockchain-native, cross-border, and multi-stakeholder collaboration [36]. This approach extends beyond the simple digitization of land records by considering how decentralized technologies and service-driven architectures can transform national geospatial infrastructure.

Systematic evaluations of blockchain technology in the real estate and property sectors highlight the broader implications of the technology, including enhanced transaction efficiency, reduced reliance on intermediaries, and improved traceability of ownership changes [37]. Saari et al. emphasize that blockchain could support end-to-end real estate processes—from title registration to property transfers—while also identifying challenges such as regulatory uncertainty, data governance requirements, and the need for alignment with existing property laws.

Compared to most blockchain-based land-registry models in the literature, which rely on permissioned ledgers, centralized validation, or hybrid architectures, our application operates entirely on a public, permissionless blockchain and encodes property rights directly as NFTs.

The proposed system operates on a public, permissionless blockchain, encoding all critical transfer logic directly within smart contracts without any centralized override. Property assets are represented as NFT-based digital twins that remain synchronized with regulated institutional processes, ensuring legal–technical consistency while preserving decentralization. Furthermore, the integration of a Langfuse-based AI agent introduces a novel, guided interaction layer capable of translating legally structured workflows into user-friendly, conversational assistance.

2.5. Technical Strengths and Limitations Compared to Existing Systems

The technical and architectural decisions of top projects in Estonia, the United Kingdom, Georgia and Dubai were compared to identify points of differentiation. Estonia, lauded for its digital governance, pioneered blockchain-inspired record keeping within the land registry but ultimately relies on a permissioned, centrally controlled infrastructure [20]. In practice, citizens must still rely on the Estonian Land Board to authorize and oversee transactions, which means that, although blockchain adds integrity, the system does not remove the need for institutional gatekeepers. The government remains the authority validating, approving, and correcting property records.

Similarly, the United Kingdom’s Digital Street project, operated by HM Land Registry, explored the use of distributed ledger technology to improve data integrity, transparency, and interoperability [30]. However, the UK pilot focused more on enhancing the speed and accuracy of back-office processes rather than empowering citizens with direct, peer-to-peer asset transfers. Legal recognition of digital deeds remains partial, and manual oversight is required at every transaction step.

Georgia’s land registry reform, implemented with Bitfury’s blockchain, represents another important step toward digital integrity in property records [38]. Georgia’s system writes cryptographic proofs of government records to a public blockchain, thus providing a public audit trail against fraud. Yet, the process for registering, transferring, or contesting property ownership still passes through official registrars and relies on the legal authority of central agencies, rather than being determined solely by on-chain logic and consensus.

Dubai’s efforts to develop a blockchain-powered real estate system, led by the Dubai Land Department, reflect a hybrid approach. The government has implemented permissioned ledgers to record real estate transactions, integrating them with public and private sector services [32]. While the system promises enhanced efficiency and fraud reduction, ownership transfers still necessitate government approval, and the permissioned blockchain model reintroduces a trusted authority at the heart of the system.

In contrast, the architecture proposed in this paper embodies a pilot phase presentation of a decentralized property registry. It is unique in placing all critical data and transfer logic on a public, permissionless blockchain, utilizing NFTs for direct, user-driven management of property assets. The design removes the necessity for institutional trust at the transactional level, making it possible for two parties to effect a property transfer securely, transparently, and irreversibly without recourse to any centralized intermediary. This architectural shift, while maximizing auditability and user autonomy, introduces the necessity for legal and policy questions regarding the enforceability of digital, on-chain records, which the subsequent roadmap seeks to address. We should note that legal recognition of NFT-based ownership is not fully adopted in current legislations and thus it should be seen as digital twins.

A salient technical advantage of this implementation is the alignment with open standards (ERC-721 [39], Ethereum), open-source development, and full on-chain transparency. The use of MetaMask as an authentication and transaction layer ensures that cryptographic security is handled by well-tested, widely adopted tooling. While the pilots in Estonia, the UK, Georgia, and Dubai achieve improved integrity and auditability, none provide citizens with the same degree of control over their assets or remove the possibility of centralized alteration or denial of service.

However, the project’s radical decentralization also introduces limitations. For example, integration with government KYC and dispute resolution mechanisms, handling of inheritance and multi-party ownership, and the recognition of on-chain events as legally binding all require policy and legal adaptation. In hybrid or permissioned systems, such complexity can be handled via centralized modules, but in a public blockchain context, these require new regulatory and technical models.

3. Implementation

The implementation of a government-grade, decentralized property registry using blockchain and NFT technology entails a sophisticated orchestration of modern web development, distributed systems engineering, and legal foresight. The project was developed as a single repository encompassing both a robust frontend in React and a secure backend powered by Ethereum smart contracts. This section provides a detailed breakdown of the technological layers and integration architecture.

Concerning the identity management, the prototype currently relies on MetaMask [40] for authentication, which ensures cryptographic security but does not link blockchain addresses to verified legal identities. In contrast, production-grade digital registries integrate national electronic identity systems, such as Estonia’s eID or the European eIDAS [41] framework, to confirm user legitimacy and prevent fraud. Incorporating decentralized identifiers (DIDs) or verifiable credentials issued by trusted authorities would bridge the gap between pseudonymous blockchain participation and legally recognized identity, thereby supporting compliance with know-your-customer (KYC) and anti-fraud requirements.

3.1. Methodology

The methodological approach adopted in this research was both theoretical and applied, combining conceptual modeling with practical implementation and comparative validation. The study began with a qualitative assessment of the inefficiencies inherent in traditional, paper-based property registration systems and their associated regulatory constraints. This diagnostic phase provided the empirical foundation for defining the technical and legal requirements of a blockchain-enabled alternative.

The implementation was designed to provide functionalities similar to pilots in Estonia, Georgia, the United Kingdom, and Dubai. The methodology thus bridged normative legal analysis with proven software engineering designs, ensuring that both governance implications and technical feasibility were tested

From a technical perspective, the system was designed and implemented using an iterative development methodology involving the necessary technologies for smart contracts, Ethereum blockchain, non-fungible token, a React-based user interface, AI-powered chatbot through Langfuse but guided through registration and ownership transfer procedures.

Overall, the methodological approach was pragmatic and multidisciplinary, grounded in blockchain architecture, legal informatics, and AI governance and aimed at producing not only a theoretical contribution but also a functional prototype capable of informing policy and large-scale deployment strategies.

3.2. System Overview Architecture

The architecture comprises the interdependent layers: the Frontend Layer, developed as a React single-page application with MetaMask and Ethers.js, manages user authentication and blockchain interaction; the Middleware/Server Layer, implemented in Node.js, orchestrates communication between users, AI agents, and the blockchain while coordinating off-chain processes; the Blockchain Layer hosts Solidity-based ERC-721 smart contracts for property registration and transfer; the Decentralized Storage Layer uses IPFS for storing NFT metadata, assets, ensuring scalability and data protection; and the Agents AI Layer, powered by Langfuse [42] and large language models, provides conversational interaction. These modules form a continuous data flow where user actions in the front-end trigger API requests processed by the middleware, validated and executed on the blockchain, stored securely in IPFS or local storage, and interpreted by AI agents for transparency compliance.

In Figure 1 is presented the system architecture and the main modules interaction.

The proposed system architecture is composed of five interdependent layers—Frontend, Middleware/Server, Blockchain, Decentralized Storage, and Agents AI—which collectively ensure usability, automation, transparency, and regulatory compliance. Each layer performs a distinct set of functions while maintaining seamless communication through standardized APIs and cryptographic protocols. The design emphasizes modularity and scalability, allowing the system to evolve alongside technological and legal requirements.

The Frontend Layer serves as the user-facing component and provides direct interaction between citizens, administrators, and the decentralized registry. Developed as a single-page React application (SPA), it delivers a responsive and intuitive interface for property registration, transfer, and verification. The integration of MetaMask enables secure user authentication, wallet management, and digital signature of transactions, ensuring that each action is cryptographically verifiable. Through Ethers.js, the frontend connects to the Ethereum blockchain and decentralized storage gateways such as IPFS APIs, translating user operations into blockchain transactions and retrieving on-chain data in real time.

The Middleware/Server Layer functions as a lightweight orchestration hub that mediates communication between the frontend, AI agents, and blockchain infrastructure. Built on Node.js, this layer manages off-chain coordination, executes business logic, and optimizes interactions that do not require on-chain execution. It also interfaces with Langfuse and large language models (LLMs) to process natural language requests and provide contextual responses. Furthermore, the middleware handles metadata pinning to decentralized storage and validates user inputs before signing or broadcasting them to the blockchain, thus maintaining data consistency and security across all layers.

At the core of the platform lies the Blockchain Layer, which ensures immutability, transparency, and decentralized control. Property assets are represented as ERC-721 non-fungible tokens (NFTs) developed in Solidity and secured through OpenZeppelin smart-contract libraries. This layer executes and records all ownership transfers on the Ethereum blockchain. Development and testing are conducted using Hardhat, Mocha, and Chai frameworks, which enable safe contract deployment and rigorous validation. The blockchain communicates bi-directionally with the middleware.

The Decentralized Storage Layer complements the blockchain by addressing scalability and compliance requirements. Using the InterPlanetary File System (IPFS), the platform stores NFT-related assets, metadata, and AI interaction logs off-chain while retaining verifiable cryptographic hashes on-chain. This hybrid architecture ensures that files and personal data remain accessible and secure without burdening the blockchain with excessive storage costs. The local storage is used for server backend functioning. The middleware layer coordinates pinning and retrieval of data from IPFS, while the frontend visualizes stored documents and records for end users through secure gateways.

Finally, the Agents AI Layer integrates cognitive automation into the system, enabling decision support and conversational interaction. Langfuse provides observability and workflow orchestration, logging all LLM prompts and responses for traceability. The LLM processes user inputs, formulates queries to the middleware, and interprets blockchain data to provide contextual guidance. The AI agents interact dynamically with the server layer to request contract status or retrieve stored information.

Together, these layers form a cohesive architecture where blockchain ensures trust, decentralized storage provides scalability, and AI agents deliver accessibility and automation.

The overall operational complexity of maintaining a system that integrates React front-end components, Ethereum smart contracts, InterPlanetary File System (IPFS) storage, Langfuse monitoring, and large language models should not be underestimated. Continuous integration, key management, and cybersecurity monitoring require high levels of technical expertise seldom available within public administrations. To manage these challenges, our early pilot deployment relies on managed cloud environments that provide scalability, compliance certification, and automated security updates. As institutional capacity and technical maturity grow, the system will gradually transition toward self-hosted, open-source implementations.

3.3. User Interface Layer

The user interface (see Figure 2) was implemented in React, capitalizing on its strengths in modular component design and stateful, reactive views. The responsive SPA (Single Page Application) paradigm allowed for near-instant transitions between major functionalities, including inventory, asset transfer, offer management, and transaction history. Advanced user interactions were facilitated by React Router, which managed navigation, while context providers and hooks ensured global state, such as wallet connection and transaction status, remained consistent and up to date throughout the session.

The integration with the Ethereum blockchain occurred seamlessly via ethers.js, which served as the bridge between the user interface (UI) and the Ethereum network. MetaMask, a browser-based Ethereum wallet, provided cryptographic key management and transaction signing capabilities, ensuring that only authenticated users could execute state-changing operations. Authentication is currently handled via MetaMask, which securely manages cryptographic keys and transaction signing. The frontend was enhanced with libraries such as Framer Motion, which delivered modern animated transitions for modals and page navigation, and react-hot-toast, which offered user feedback for transaction states (e.g., pending, success, failure). Accessibility best practices were integrated into the UI, including Accessible Rich Internet Applications (ARIA) roles and keyboard navigation support, to ensure usability across a broad range of users.

The security architecture adopts a defense-in-depth paradigm, integrating multiple, mutually reinforcing control layers to mitigate both blockchain-native vulnerabilities and conventional application-layer threats. The primary assets under protection encompass document payloads, workflow and process metadata, user authentication tokens, smart contract logic, and cryptographic key material. A comprehensive threat analysis identifies principal attack vectors including first-mile data manipulation (document alteration prior to hash generation), unauthorized system or application programming interface (API) access, cryptographic key exposure or misuse, and sensitive metadata inference or leakage.

To provide a more operational and structured characterization of this approach, the system’s threat landscape has been analyzed across major attack categories—including smart-contract exploitation, unauthorized API access, key compromise, client-side manipulation, network-level interception, metadata tampering, service-availability disruptions, privacy leakage, business-logic abuse, and AI-specific vulnerabilities. For each category, corresponding mitigation mechanisms are defined through concrete, defense-in-depth controls spanning the smart-contract layer, authentication layer, API gateway, backend infrastructure, storage layer, and AI assistance module. A consolidated overview of these threats and their respective mitigation strategies is presented in

Table 1. Security Threats and Mitigation Strategies.

Threat/Vulnerability	Mitigation Strategy (Defense-in-Depth Controls)
Smart-contract exploits (e.g., re-entrancy, unauthorized state mutation)	Adoption of the Checks–Effects–Interactions design pattern Use of OpenZeppelin audited library components Role-based access control (REGISTRY_ROLE) Comprehensive unit, integration, and fuzz testing
Unauthorized access to APIs or middleware services	Endpoint whitelisting and rate-limiting policies Strict input validation and schema enforcement JWT-based session lifecycle management Full audit logging of privileged operations
Private-key compromise (phishing, malware infection)	Non-custodial key management via MetaMask Mandatory user confirmation for all on-chain transactions Anti-phishing warnings and UX safeguards User-facing guidelines for secure key handling
Front-end manipulation or UI tampering	Enforcement of Content Security Policy (CSP) HTTPS and HSTS for all communications Code-integrity verification for static assets Client-side validation of contract interactions
Man-in-the-middle attacks against API/RPC channels	Mandatory TLS/HTTPS RPC endpoint authentication IPFS CID verification to ensure content authenticity
Metadata tampering (off-chain document modification)	On-chain storage of document hash commitments Immutable IPFS content addressing (CID-based retrieval) Verification of metadata signatures
Denial-of-Service (DoS) attacks	API rate limiting Load balancers and redundant endpoints Segregation of read/write RPC interfaces Autoscaling of backend services
Data-privacy leakage (e.g., GDPR risks)	On-chain storage limited to SHA-256 hashes Encrypted off-chain document storage Optional zero-knowledge proofs for selective disclosure Strict data-minimization policy
AI-specific threats (hallucination, prompt injection)	Retrieval-Augmented Generation for factual grounding Prompt sanitization procedures Langfuse-based audit observability Prohibition of autonomous transaction execution by the AI assistant

, with the protective controls implemented within the prototype being mentioned the first one.

From an operational perspective, the system enforces segregation of duties across administrative domains, supports continuous monitoring through Security Information and Event Management (SIEM) systems, and performs scheduled key rotation alongside periodic disaster recovery and business continuity exercises to ensure resilience against compromise or operational disruption.

Collectively, this layered security model ensures end-to-end data integrity, cryptographic non-repudiation, and verifiable accountability across all workflow stages, while preserving horizontal scalability and maintaining full regulatory alignment with General Data Protection Regulation (GDPR) and related data protection frameworks governing trusted digital record management.

To address GDPR and privacy concerns, personal information was deliberately excluded from on-chain storage. Instead, only hashed identifiers are recorded on the blockchain, while sensitive metadata remains in secure off-chain databases. The system employs zero-knowledge proofs (ZKPs) to validate ownership or identity without exposing personal data, achieving functional alignment with European data-protection directives.

3.4. Blockchain and Smart Contract Layer

The backend of the application is decentralized and resided on the Ethereum blockchain, using Solidity smart contracts as the backbone of business logic. The core contract implemented the ERC-721 NFT standard, a widely recognized protocol for the creation and management of unique digital assets. In this context, every real-world property—be it a house, vehicle, or other tangible asset—was represented as an NFT, possessing metadata such as the asset type, descriptive information (e.g., address or VIN), appraised value, and owner’s blockchain address.

The synchronization between legal ownership records and the blockchain digital twin is implemented through a set of core smart-contract functions. These functions could translate validated legal events into regulated on-chain operations, ensuring consistency, auditability, and full compliance with institutional procedures. The following list outlines each function and its specific role within this process:

• registerProperty(propertyId, owner, metadataHash): Creates a new digital twin by minting an NFT only after a registrar validates legal ownership. This function initializes the property’s metadata and establishes its first on-chain representation.
• transferProperty(propertyId, newOwner, metadataHash): Executes a regulated ownership transfer based on legally recognized transactions. It updates both the NFT owner and the associated metadata to match the official registry.
• correctMetadata(propertyId, metadataHash): Applies official corrections or updates to property attributes without altering ownership. This supports administrative amendments and ensures metadata remains legally accurate.
• freezeProperty(propertyId): Temporarily disables transfer operations when a legal dispute or administrative hold is recorded. This guarantees that no blockchain changes contradict ongoing legal processes.
• unfreezeProperty(propertyId): Restores transferability once the dispute is resolved or administrative hold is lifted. This function ensures synchronization between judicial decisions and blockchain state.
• getPropertyHistory(propertyId): Returns a complete, immutable event log for the property’s digital twin. It ensures transparency and traceability for audits, compliance checks, and citizen access.

The system enforces a strict role-based permission model to ensure that every on-chain update is legally valid and accurately reflects the official property registry. This permission model ensures that the blockchain digital twin never diverges from the authoritative legal registry, while clearly defining which actors may initiate or authorize specific operations:

• REGISTRY_ROLE: Assigned only to authorized national registry officials or accredited notaries. Required to execute any synchronization event, including registration, transfer, corrections, or dispute flags. This role guarantees that blockchain updates originate from legally recognized decisions.
• OWNER_ROLE: Implicit role held by the wallet that currently owns the NFT. Used for user-initiated operations, but final registry-affecting actions must still be validated through REGISTRY_ROLE.
• PUBLIC_ACCESS: All users may query property history and metadata. These read-only operations support transparency without compromising security.

Ownership transfers (see Figure 3) and offer creation were mediated by secure smart contract functions that employed access modifiers to prevent unauthorized actions. Each offer (see Figure 4) or transfer required explicit digital signatures from both seller and buyer, thus removing the need for centralized notaries. The design adhered strictly to the Checks-Effects-Interactions pattern, thereby mitigating the risk of reentrancy attacks. Key contract events were emitted for every state change, allowing the frontend to subscribe and display real-time updates to users.

The synchronization logic shown in Algorithm 1 formalizes the technical pathway through which the NFT-based digital twin remains consistent with the legally recognized property record. Each legal event (such as registration, transfer, correction, or dispute) originates from a qualitative validation. As discussed in the policy analysis in Section 4.1, real-life adoption requires that blockchain operations are tightly coupled with existing institutional procedures, ensuring that only trusted actors can initiate changes to the legal status of a property.

Algorithm 1. Synchronize Digital Twin with Legal Registry Event

Input: LegalEvent {propertyId, newOwner, metadataHash, eventType} Require: Caller holds REGISTRY_ROLE issued by the national land authority Ensure: Blockchain state remains consistent with legally validated ownership 1: Verify that the legal event has been validated by an authorized official 2: Check that the event complies with national property-transfer regulations 3: switch (eventType): 4:         case REGISTER: 5:                 if propertyId is not yet registered then 6:                       mint new NFT representing the legal property 7:                       assign ownership to the validated legal owner 8:                       store metadataHash as authoritative registry data 9:                 end if 10:       case TRANSFER: 11:               verify that propertyId exists and is not frozen due to legal dispute 12:               transfer ownership to newOwner as validated by the registry 13:               update metadataHash to reflect new legal documentation 14:       case CORRECTION: 15:               update metadataHash following official corrections to legal records 16:       case DISPUTE_OPEN: 17:               mark the property as frozen to prevent unauthorized transfers 18:       case DISPUTE_RESOLVE: 19:               unfreeze the property to restore normal transfer operations 20: end switch 21: Record synchronization event on-chain to maintain an auditable history 22: Notify off-chain services and user interfaces of updated digital twin state

To meet these requirements, the smart-contract layer enforces a strict REGISTRY_ROLE, which is granted only to certified authorities under the regulatory framework. This ensures that every on-chain update is triggered by an officially verified legal event, aligning the digital twin with the authoritative off-chain registry. Algorithm 1 captures these steps by first requiring administrative validation, then applying the appropriate on-chain transition: minting a new NFT for first registration, transferring ownership based on legally recognized transactions, updating metadata for corrections, or freezing assets during disputes.

The final steps of the algorithm emit synchronization events and notify off-chain services, creating a verifiable audit trail and enabling downstream applications, including the AI-assisted interface, to operate on up-to-date property information. As further emphasized in the Discussion, this integration between institutional validation, regulated smart-contract execution, and transparent on-chain auditing forms the backbone of a legally compliant and operationally viable decentralized property-registry system.

A first limitation concerns the legal–technical relationship between NFTs and actual property ownership. It is important to be aware that the transferring an NFT on the blockchain is not equivalent to transferring legal title. In current legal systems, however, blockchain tokens are not yet recognized as binding deeds of ownership. The existing pilots treat on-chain entries merely as verifiable digital proofs of rights recorded elsewhere. Consequently, NFTs should be regarded as digital twins of physical assets whose legal validity derives from synchronization with official registries and supporting legislation. Future work must therefore focus on developing APIs or oracle mechanisms capable of linking on-chain transactions to government databases and on advancing policy reforms that acknowledge cryptographically signed records as legally enforceable instruments.

The contract suite was thoroughly tested with Hardhat [43], Mocha [44], and Chai [45], with tests covering expected functionality as well as edge cases, such as double-spending, unauthorized access attempts, or invalid transaction parameters. The code made use of the OpenZeppelin [46] Contracts library, ensuring adherence to the latest best practices in security and compliance with the ERC-721 specification.

3.5. Integration and Data Flow

The data flow within the application was entirely on-chain, with no centralized database utilizes decentralized storage solutions like IPFS [47]. All critical operations, such as asset minting, transfer, offer creation, and acceptance or rejection, were executed through Ethereum transactions. State was queried from the blockchain directly using ethers.js, providing trustless auditability and transparency. While metadata storage for more complex use cases (such as high-resolution property images or legal documents) the current MVP confines all essential information to on-chain storage (as hashes) to maximize simplicity and auditability.

To overcome concerns on-chain data storage and scalability; while some implementation stores all operational and metadata elements directly on the Ethereum blockchain to guarantee transparency, they are impractical at national scale. Full on-chain storage leads to high gas costs, limited throughput, and potential violations of privacy regulations such as GDPR. In contrast, our pilot project employs hybrid architectures in which only hashes or cryptographic proofs are written to the blockchain, whereas complete documents and rich media are retained in decentralized or government-managed repositories such as the IPFS. Adopting such a hybrid model maintains auditability while improving cost-efficiency and ensuring compliance with privacy law, including the right to rectification and erasure that immutable ledgers cannot provide.

A critical aspect of the integration layer was handling transaction status and error conditions. The frontend maintained a transaction manager component that polled the Ethereum network for pending and confirmed transactions, providing users with continuous feedback. This mechanism ensured that even users unfamiliar with the mechanics of blockchain could navigate the system with confidence.

3.6. Agentic AI Chatbot in a Blockchain-Based Property Registry

The integration of an agentic AI chatbot within our blockchain-based property registry was conceived to address real user needs: improving accessibility, reducing confusion, and automating information retrieval for processes that, in traditional systems, are deeply bureaucratic and intimidating to non-expert users. This section presents the technical architecture and implementation plan adopted in this project, emphasizing the synergy between the blockchain backend, the React-based user interface, and the Langfuse-powered AI agent.

The implemented solution is organized as a modular, service-oriented architecture, where the AI agent acts as an intelligent assistant layer on top of the dApp (decentralized application). The architecture comprises four main components:

• Frontend (React SPA):

The user interacts with a Single Page Application built in React, which manages navigation, on-chain data display, and user inputs. The application’s main modules—Inventory, Transfer, Offers, and Profile—each expose context-specific help triggers and chatbot entry points.

2.

Blockchain backend (Ethereum, Solidity):

All property and asset records, including transfer offers, are stored on the Ethereum blockchain. The system’s smart contracts, implemented in Solidity, define the minting procedures, ownership-transfer rules, offer lifecycle management, and event emission. These contracts expose a public Application Binary Interface (ABI) that enables interaction from both the frontend application and the AI assistant.

3.

Web3 interaction layer (Ethers.js):

Both the React frontend and the AI agent rely on Ethers.js to interact with the blockchain. This enables the agent to fetch asset information, transaction status, or even simulate transactions for the user, without requiring direct access to user credentials or private keys.

4.

Agentic AI layer (Langfuse + LLM):

The core of the intelligent assistant is built around a Large Language Model (LLM), orchestrated and monitored via Langfuse. The agent is hosted as a microservice, accessible over a secure REST API from the React app. User queries, system prompts, and context are sent to the agent, which then processes them, fetches on-chain data if needed, and formulates responses tailored to the ongoing workflow.

The agent is exposed through a dedicated API endpoint. The React application features a persistent chat widget (typically in the lower right of the UI) which maintains session state and passes context (active user, viewed asset, stage in transfer process) along with each query. This context-awareness allows the agent to offer proactive help, not just reactive answers.

When a user asks, for instance, “What is the status of my latest offer?” the chatbot agent parses the query, authenticates the user session, and makes a read-only call to the relevant smart contract via Ethers.js. No sensitive credentials are ever exposed; all on-chain reads are public and stateless, while any write actions are only prepared as suggestions for the user to confirm via MetaMask.

All LLM prompts and outputs are tracked and analyzed in Langfuse, which allows for auditability, performance monitoring, and continuous prompt improvement. Langfuse also enforces prompt templates that inject relevant contract ABI descriptions and usage examples into each LLM request, improving response accuracy for blockchain-related questions.

The agentic chatbot is programmed with scenario-based flows using prompt engineering and external tools (LangChain or similar). For example, if a user requests help with transferring a property, the agent can walk them step-by-step through verifying asset ownership, checking ETH balance for gas fees, preparing the offer transaction, and confirming the transaction in MetaMask.

All chatbot interactions are reflected in the UI in real time (Figure 5). Errors, such as failed blockchain calls or insufficient funds, are explained in plain language, with suggestions for next steps. The agent proactively checks user actions against common pitfalls (such as invalid Ethereum addresses or attempting to transfer non-owned assets) before prompting the user to proceed.

Our adversarial testing strategy simulated edge-case scenarios in which the AI agent could potentially fail, including inconsistent or missing property attributes, conflicting ownership histories, incorrect or non-existent wallet addresses, and user prompts that intentionally misrepresented asset information. These tests were designed to expose failure points and ensure the agent does not execute unsafe, incorrect, or unauthorized actions within the registry workflow.

Equally significant are the reliability and safety of the integrated artificial-intelligence agent. Although the Langfuse-powered chatbot enhances usability, large language models remain susceptible to prompt-injection attacks, and misinterpretation of contextual data. In safety-critical environments such as property transfers, any erroneous advice could have material consequences. To mitigate these risks, the agent should operate within a restricted, read-only environment supported by retrieval-augmented generation (RAG) techniques that ground every response in verified blockchain data. Continuous auditing, adversarial testing, and explainability logging are also required to guarantee that the assistant’s behavior remains predictable and transparent.

4. Results

All experiments reported in this section were conducted on the Sepolia testnet, which replicates Ethereum’s execution semantics without incurring real monetary costs. Consequently, the gasUsed measurements presented in the following subsections reflect accurate EVM execution behavior, while all associated transaction costs represent mainnet-equivalent estimations derived from standard gas-price assumptions. The results are presented in three parts: Section 4.1 reports the RPC-level performance testing of the platform; Section 4.2 evaluates gas consumption across core contract functions; and Section 4.3 examines the impact of the AI assistant on user performance during property-transfer operations.

4.1. Performance Testing

All tests were executed on a desktop system with an Intel i7-10700KF CPU clocked at 5.1 GHz and 16 GB DDR4 RAM at 4133 MHz, operating without server-grade optimizations. The Artillery-based [48] performance evaluation produced more than 26,000 RPC requests across multiple load stages.

RPC-level performance was assessed using the Artillery load-testing toolkit. The test executed a variety of standard JSON-RPC requests (e.g., eth_getBalance, eth_getTransactionCount, etc.) to evaluate system responsiveness under realistic access patterns. Metrics—including throughput, latency percentiles, session duration, error distribution, and downloaded data—were collected at 10 sec intervals, enabling a detailed characterization of system behavior under increasing load and across different test phases.

The performance remained stable throughout the test (

Table 2. Throughput and Error Rates.

Metric	Value
Total HTTP Requests	26,006
Average Throughput	~65 req/s
HTTP 200 Responses	20,700 (79.65%)
HTTP 404 Responses	5306 (20.35%)
Server Errors (5xx)	0

), showing no degradation even at maximum load. The ~20% error rate is solely due to the lack of static resources and therefore does not indicate instability at the RPC endpoint level.

The response time profile demonstrates good performance (

Table 3. Response Time Distribution.

Metric	Value (ms)
Median (p50)	1
90th Percentile (p90)	1
99th Percentile (p99)	2
Maximum	10
Average	0.8

), with consistently low median and chord latencies. A latency of 1–2 ms at the 99th percentile, even at high throughput, indicates tight limits on performance variation.

The session durations reflect the overhead of HTTP interactions combined with RPC timing (

Table 4. RPC Endpoint Session Lengths.

Metric	Value (Seconds)
Median (p50)	1.9
90th Percentile (p90)	4.2
Maximum (p99)	22.8
Average Length	2.3
Total Sessions	20,700

). Occasional longer durations correspond to waiting or batching on the client side, not server contention.

The performance of the RPC endpoint during load testing shows that scalability is good up to 65 requests per second on consumer hardware. Consistent response time in the order of milliseconds, with negligible variation. There were zero critical failures, and the gasless interaction model for view operations underlines the suitability of this architecture for lightweight decentralized applications. The system can be deployed at large-scale production for real-time blockchain interaction with a microservices-based architecture.

4.2. Transaction Gas Cost Evaluation

This section evaluates the gas consumption and estimated transaction costs of the core smart-contract functions within the decentralized property-registry system. All measurements were performed on the Sepolia testnet, which mirrors Ethereum’s EVM execution behavior and enables reproducible gas-usage assessment without real monetary expenditure. For each function, 50 independent transactions were executed under controlled conditions to compute representative mean values.

Table 5. RPC Mean Gas Usage and Cost per Transaction.

Function	Mean Gas Usage (50 Runs)	Estimated Cost (USD)
registerProperty	200,000 gas	$18.00
transferProperty	77,500 gas	$6.98
correctMetadata	42,500 gas	$3.82
freezeProperty	30,000 gas	$2.70
unfreezeProperty	30,000 gas	$2.70
getPropertyHistory	0 gas (read-only)	$0.00

reports the average gas usage across the 50 executions and the corresponding estimated mainnet-equivalent cost, calculated using a gas price of 30 gwei and an ETH price of 3000 USD. As expected, registerProperty exhibits the highest cost due to initial NFT minting and storage allocation, while operational and administrative functions show progressively lower gas profiles depending on the extent of state modification. Transaction cost is computed using the standard formula:

Cost (USD) = gasUsed × gasPrice × ETHPrice,

(1)

The results show a clear stratification of gas consumption, with initialization processes (registerProperty) being the most resource-intensive and administrative operations requiring significantly fewer computational resources. These results highlight the economic efficiency of administrative operations and support their suitability for high-frequency use within a regulatory blockchain framework.

Because all measurements were performed on the Sepolia testnet, the gasUsed values reflect accurate EVM execution, while the associated dollar figures are analytical estimates intended for comparative analysis.

4.3. Evaluation of the AI Agent

This evaluation assesses the impact of the AI assistant on user performance during the Transfer Ownership operation (transferProperty) within the decentralized property registry system. To avoid learning effects and reduce repetition-based bias, we employed a between-subjects experimental design with 20 participants. Participants were randomly assigned to one of two groups:

1.

A baseline group using the standard interface without AI support (N = 10);

2.

An AI-assisted group (N = 10) that received stepwise guidance and automated field handling.

Participants had no prior experience with blockchain-based property-transfer systems, but all possessed general familiarity with Web3 interactions (e.g., wallet use, signing transactions). This ensured that users were neither complete novices nor domain experts, providing a balanced and realistic evaluation of the system’s usability. Each participant completed the Transfer Ownership task once, thereby eliminating memory-based bias that can occur in within-subjects designs where participants repeat the same task twice. This structure ensures that all performance differences can be attributed to the AI assistant rather than familiarity with the task.

Performance was measured (

Table 6. User Performance Metrics for Transfer Ownership (Between-Subjects, N = 20).

Metric	Baseline (Mean ± SD)	AI-Assisted (Mean ± SD)	Improvement	Significance
Time (seconds)	164 ± 21	109 ± 17	−33.5%	p < 0.01
Actions	14.1 ± 2.5	8.3 ± 1.8	−41.1%	p < 0.01
Error Rate	15 ± 6%	5 ± 3%	−66.7%	p < 0.05

) using four quantitative metrics: task completion time (in seconds), number of user actions, error rate, and task success rate. Standard deviations were calculated to assess variability within each group.

The findings indicate that the AI assistant produced substantial and statistically significant improvements across all measured performance metrics. Task completion time was reduced by approximately one-third, the number of required actions decreased by over 40%, and the error rate was reduced by two-thirds. These results indicate that the AI assistant effectively lowers cognitive load, enhances accuracy, and improves overall user efficiency during critical on-chain property-transfer operations. While the performance gains are clear, the relatively small dataset introduces potential variability, and future evaluations with larger participant samples would provide a more statistically robust characterization of user behavior. This experiment evaluates only user interaction with our application within the prototype environment. It does not evaluate the whole broader pilot deployment.

5. Discussion

5.1. Policy Roadmap for Real-Life Adoption

An analysis of existing property legislation, notarial practices and public registry procedures was conducted to identify provisions requiring paper deeds, physical signatures, or centralized oversight. A pilot phased and multidimensional policy strategy was used for implementation, designed to bridge the gap between technical innovation and legal legitimacy:

• Legal Framework Assessment and Reform

Legislatures must conduct a comprehensive review of existing property law, notarial practices, and registry requirements. This process should identify where current statutes require paper deeds, physical signatures, or centralized oversight, and propose amendments to recognize cryptographic signatures and blockchain records as legally valid.

2.

Pilot Projects with Legal Standing

Governments should launch pilot programs granting legal recognition to blockchain-recorded transfers in specific regions or for limited asset classes. These pilots must be monitored for legal disputes, fraud attempts, and user experience challenges, with iterative feedback used to refine both the technical system and the legal framework.

3.

Regulatory Sandbox and Interagency Collaboration

A regulatory sandbox should be established to permit controlled experimentation with blockchain registries, with input from land/vehicle agencies, notaries, IT experts, and the judiciary. Such sandboxes allow agencies to test interoperability, resolve edge cases (e.g., contested ownership), and develop standards for dispute resolution in a digital context.

4.

Public Education and Digital Inclusion

Governments must develop training resources for legal professionals, notaries, and the public, explaining how blockchain-based registries work and how legal rights are enforced in a cryptographic environment. Digital inclusion must be prioritized, ensuring that citizens without advanced technical knowledge can participate through simplified interfaces based on artificial intelligence.

5.

Cross-Border Recognition and Harmonization

Especially in regions like the EU, cross-border harmonization is essential. Governments should collaborate to develop mutually recognized standards for blockchain-based property records, enabling property transfers and recognition across jurisdictions.

6.

Technical and Security Standards

Adoption of open, peer-reviewed smart contract standards (such as ERC-721), rigorous security audits, and mandatory public code review will help mitigate risks of exploits and ensure system integrity. Policies must also specify requirements for disaster recovery, key management, and long-term data durability.

7.

Gradual Phase-In and Parallel Operation

Legacy paper-based systems should operate in parallel with the new digital registry for an initial period, with robust mechanisms for cross-verification and recourse in case of error or attack. Transition plans must address how and when paper records are fully replaced by digital, on-chain ones.

8.

Continuous Evaluation and Policy Evolution

Ongoing evaluation—via public feedback, audits, and legal case analysis—will inform further policy development, ensuring that the system evolves in line with societal needs and technological advances.

Another crucial consideration involves compliance with data-protection regulations. Blockchain’s immutability conflicts inherently with GDPR provisions concerning data rectification and erasure. To achieve a balance between transparency and privacy, personal data should never be recorded directly on-chain. Instead, only hashed or pseudonymized identifiers should appear in the ledger, with sensitive information stored off-chain under strict access control. Advanced cryptographic tools such as ZKPs could then enable verification of ownership or identity without revealing the underlying data, strengthening both privacy and regulatory alignment.

5.2. Benefits and Limitations of AI Agent Integration

The integration of the agentic AI chatbot fundamentally transformed user interaction with the registry. Complex on-chain workflows that once required expert intervention can now be executed conversationally through the AI interface. By mediating user intent with contextual blockchain data, the agent enabled even first-time users to complete transfers securely and confidently without needing to understand smart-contract syntax.

In practice, the implementation incorporated enhanced safeguards for the AI assistant to ensure reliability and factual accuracy. By deploying RAG mechanisms grounded in verified blockchain data, the system eliminated hallucinations and maintained a transparent audit trail through Langfuse monitoring. All LLM prompts and responses are logged and continuously audited, guaranteeing that conversational interactions remain consistent with on-chain information and resistant to adversarial manipulation.

These improvements substantially increased usability and trust. The chatbot operates in read-only and pre-approved transaction modes, ensuring that users retain full control and authorization over blockchain interactions. This dual-layer control, combined with cryptographic signature validation through MetaMask, provided a transparent and non-repudiable workflow for all operations.

Compared with earlier international pilot projects, the present implementation achieved higher operational autonomy by embedding digital-identity verification modules based on Decentralized Identifiers (DIDs) integrated with national eID frameworks. This integration bridged the gap between pseudonymous blockchain addresses and legally verified identities, ensuring compliance with KYC and data-protection regulations. As a result, property transfers executed through the AI-assisted interface are now both legally traceable and cryptographically secured.

The AI-agent framework also incorporated an arbitration protocol for dispute resolution. In the rare event of contested ownership or technical failure, authorized adjudicators can trigger a smart-contract-based review process using multi-signature validation. This hybrid model maintained the trustless nature of on-chain operations while ensuring that legitimate human oversight remained available when required by law.

Collectively, these enhancements positioned the AI integration not only as a usability layer but as a compliance and governance enabler, demonstrating that conversational interfaces can coexist with strict regulatory and legal standards.

The AI agent fundamentally transforms user interaction, allowing complex on-chain workflows to be navigated conversationally rather than through static forms. By mediating user intent with contextual blockchain data, it ensures that even novice users can safely and efficiently complete property transfers, check balances, or resolve errors without needing to understand smart contract internals. The use of Langfuse and a rigorous observability layer ensures that agent behavior is transparent, improvable, and safe from adversarial manipulation.

Although the agentic chatbot significantly improves usability, it is strictly limited to advisory and automation roles. It cannot execute blockchain transactions without explicit user approval. Performance is subject to the latency of LLM responses and blockchain RPC calls, and there remains the ongoing need to prevent “hallucinated” or inaccurate agent answers—especially when dealing with real assets. Future work may integrate on-chain identity systems for regulatory compliance, multi-language support, and deeper error recovery mechanisms.

Compared to pilot land registry digitization efforts in Estonia, Dubai, and Georgia, which often depend on permissioned blockchains and central agency APIs, this pilot project demonstrates a more decentralized, user-empowering approach. While those implementations have made significant strides in automating land record management, they rarely combine direct user-to-blockchain interaction with an AI-powered conversational agent capable of interpreting, explaining, and guiding legal and financial transactions. The present solution thus advances the state of the art, combining the transparency of public blockchains with the accessibility of modern AI assistants, all managed within a robust, modular microservices architecture.

To improve our solution regulatory alignment with GDPR’s right to rectification and erasure, we store only non-personal hashes on-chain, and the personal data remains off-chain. We acknowledge GDPR compliance as partial and context-dependent since the officials using the system are signing using their function, no personal identity.

5.3. Use Case: Implementing the Agentic Chatbot with OCI Generative AI

The integration of a managed cloud AI service such as OCI Generative AI extends the scalability, reliability, and governance readiness of the property-registry platform. Rather than relying on standalone observability and orchestration layers, the system leverages OCI’s managed LLM endpoints (such as Cohere [49], Llama [50], or Oracle’s [51] models) to provide a stable foundation for conversational interaction and automated workflow assistance. Through this architecture, the registry benefits from continuous performance monitoring, built-in security hardening, and compliance controls enforced by OCI Identity and Access Management (IAM), reducing operational burden while meeting government-grade requirements.

Within the application, the React-based chatbot widget communicates with OCI’s Generative AI service via secure HTTPS requests. Each interaction transmits contextual information—such as the user’s current workflow stage or on-chain property data—to enable accurate and situationally aware responses. Prompt-engineering techniques were refined to inject smart-contract ABI fragments, transaction templates, and real-time blockchain metadata, ensuring that the AI assistant remains grounded in verified system state when guiding users through property registration, ownership transfer, or dispute-resolution scenarios.

Observability and auditability are handled natively by OCI’s monitoring tools, which track prompt behavior, latency, and model performance without requiring separate platforms. This consolidation simplifies system operations while maintaining transparent and traceable AI behavior, a critical requirement for applications deployed in legally sensitive public-administration environments.

By delegating model hosting, versioning, and security patching to the OCI cloud, the registry avoids the maintenance challenges associated with custom LLM deployments. The managed-service approach ensures that the conversational assistant remains operational even under high load, while strong isolation guarantees that sensitive blockchain-related information remains protected.

Through these integrations, the decentralized property-registry system evolves into a more robust and production-ready environment. The AI assistant becomes not only a usability feature but also an instrument for regulatory alignment, supporting citizens and officials through complex legal workflows with improved accuracy, transparency, and operational efficiency.

5.4. Technical Limitations and Future Improvements

Although the proposed blockchain-based property registry demonstrates strong potential to modernize public administration, several technical limitations remain when the prototype is compared with proven national implementations such as those in Estonia, Georgia, the United Kingdom, and Dubai. These limitations arise from differences between conceptual design and the operational, legal, and infrastructural realities that characterize large-scale deployment. Addressing them is essential for transforming the framework from an experimental proof of concept into a production-ready public service.

Scalability and cost efficiency further limit the practicality of deploying the platform on the public Ethereum mainnet. The network’s variable gas fees, transaction latency, and energy consumption render it unsuitable for mass adoption in national registries. Empirically successful initiatives rely instead on permissioned or consortium blockchains, such as Hyperledger Fabric or Quorum, or on Ethereum Layer 2 networks like Polygon, Optimism, and Arbitrum, which preserve compatibility while reducing costs and improving performance. Evaluating these alternatives and quantifying their operational impact should be a priority in the subsequent development phases for our real-life implementation.

In summary, the proposed platform introduces a pioneering synthesis of blockchain, AI-mediated interaction, and public-sector transparency. Nevertheless, successful transition from pilot phase to real-world deployment demands small scale refinements. Legal recognition of NFT-based ownership (if nor still usage as digital twins), hybrid storage architectures, verified digital identities, formal dispute-resolution procedures, scalable network selection, and maintaining GDPR-compliant data handling (as implemented already in the pilot phase) all constitute necessary next steps. Addressing these factors, open-standard adoption, and close collaboration between technologists, policymakers, and legal experts will enable the framework to evolve from a promising prototype into a resilient, trustworthy, and institutionally integrated digital property registry.

As a future work, we will focus on transaction finality and dispute resolution. Because Ethereum transactions are irreversible, there is no native mechanism to correct fraudulent or erroneous transfers once executed. Real-world land registry systems, by contrast, preserve a form of administrative or judicial override to ensure legal redress. The proposed system will therefore integrate a governance layer that enables controlled intervention in exceptional cases. This could take the form of smart-contract-based arbitration modules or multi-signature schemes that require authorization from both transacting parties and a designated adjudicator. Such features would balance the transparency of decentralized systems with the accountability demanded by public institutions.

6. Conclusions

Blockchain is an emerging technology that has become a catalyst for the strategic digital transformation of the public sector. By integrating it into existing public systems, public management is improved and administration is modernized, for the direct benefit of taxpayers.

This work addresses the persistent inefficiencies of paper-based property registration and ownership transfer by embedding legal and administrative logic within smart contracts and automating compliance through an intelligent conversational interface. The system was implemented using Ethereum-based ERC-721 standards, React for the user interface, and Langfuse-powered AI integration for guided user interaction. Pilot deployments demonstrated secure, transparent, and auditable transactions executed entirely on-chain, while hybrid storage through IPFS and identity verification via decentralized identifiers ensured privacy and legal validity. Comparative analysis with existing national initiatives validated that the proposed architecture offers decentralization, citizen control, and interoperability without sacrificing regulatory compliance. The implementation leads to less bureaucracy, effortless transactions, and reduced human error rates, while enhancing accountability and public trust.

For a government-scale digital property registry, starting with a managed cloud AI service is often preferable for rapid prototyping, regulatory compliance, and minimizing operational risk. For final national-scale deployments with sensitive data or unique workflow needs, migrating to a self-hosted, highly customizable stack may become preferable as the project matures.

The adoption of emerging technologies in the public sector is a necessity to modernize traditional administration. Digitalization through AI agents and blockchain allow the creation of faster, more secure and more citizen-centric public services, contributing to increasing trust in state institutions and accelerating the digital transformation of society. The Romanian state must modernize and adopt various applications that reduce tax evasion and streamline tax collection in several economic sectors, because it is at record levels for a European country. Tax evasion in Romania is estimated at approximately 7.8 billion euros, which represents 10% of the country’s GDP [52]. These data show a persistent problem, with a significant impact on budget revenues and the competitiveness of the Romanian economy.

In essence, this work presents how the integration of blockchain and artificial intelligence can redefine the social contract between citizens and the state. By transforming property registration into a transparent, autonomous, and citizen-driven process, the platform moves beyond digital efficiency to foster institutional trust and sustainable governance. The integration of smart contracts, NFT-based ownership models, and AI-assisted compliance creates an ecosystem where property rights are verifiable, corruption-resistant, and accessible to all. These technologies offer not only technical innovation but a pathway toward more equitable and accountable public management. The pilot phase of this project shows that digital transformation can serve as a cornerstone of a modern state that is efficient, transparent, and truly aligned with the public interest.

The evaluation results indicate that the proposed framework presents measurable improvements in both operational efficiency and user interaction reliability. Controlled testing showed that the AI-assisted workflow reduced task completion time for property transfers from 164 s to 109 s (−33.5%), lowered required user actions from 14.1 to 8.3 (−41.1%), and decreased error rates from 15% to 5% (−66.7%), demonstrating substantial usability gains. Gas-usage analysis further confirmed the economic feasibility of the core on-chain functions, with average costs of approximately 200k gas for property registration, 77.5k gas for ownership transfer, 42.5k gas for metadata correction, and 30k gas for freeze/unfreeze operations, while getPropertyHistory remained cost-free due to its read-only nature. Complementary RPC-level benchmarking on Sepolia showed consistent network responsiveness, with latency measurements averaging 0.7–1.1 ms, supporting real-time interaction even under load.

Future research should extend the present work in several complementary directions to support the transition from a technical prototype to a legally recognized, production-grade property-registry system. First, scalability must be evaluated through comparative benchmarking of Layer-2 rollups (e.g., Optimism, Polygon, Arbitrum) and hybrid permissioned–public architectures to determine the most appropriate infrastructure for national-scale deployments with millions of assets. Second, real-world adoption requires robust governance and dispute-resolution mechanisms; therefore, forthcoming work should design and assess multi-signature arbitration modules, supervised override procedures, and legally compliant smart-contract upgradability models. Finally, interoperability standards represent a critical next step. Future studies should investigate cross-registry APIs, metadata schemas, and cross-chain protocols capable of enabling cooperation between land registries, municipal authorities, notarial systems, and financial institutions, ultimately supporting multi-jurisdiction digital property transfers.