The Decentralized Autonomous Organizations (DAOs) are shaping the future of the governance by moving away toward the power of the communities making calls without a central body. Nevertheless, it is becoming harder to assess the quality of the proposals as those are increasing and the number of demands is increasing as well. Manual reviews need more man power, lack consistency and are prone to bias because each one can produce varying levels of clarity, possibility, and fit within organizational objectives. In this paper, we introduce the proposal evaluation system based on AI, which uses transformer-based Natural Language Processing (NLP) models and Explainable AI (XAI) to automate and interpret the assessments of DAO proposals. The system scores in three dimensions, including impact, feasibility, and goal alignment in a clear and continuous way, with the justifications in human-readable formats. Our solution promotes both the fairness and scalability of decision-making in DAOs by decreasing voter fatigue and achieving a more straightforward workflow in governing the activities. Trained and validated on real-world DAO proposal datasets, the model delivers high performance regarding accuracy, explainability, and user trust. The contribution of this project is to guide the community to the intelligent systems of governance in Web3 through how to increase the transparency and trust in them using automated decision-support tools leaving the decentralized nature of DAOs intact.
Recent advances in large language models (LLMs) have enabled the emergence of intelligent agents capable of performing complex multi-step tasks across various domains. In parallel, the growth of Web3 has introduced a decentralized web infrastructure, yet remains largely inaccessible to non-technical users due to operational complexity, fragmented information, and security risks. In this article, we present Web3Agent , an AI agent system that integrates LLM-based interaction with blockchain environments to enable language-driven on-chain operations. Web3Agent automatically decomposes user instructions into structured workflows, dynamically queries blockchain data and APIs, and performs multi-step operations such as asset transfers, token swaps, and smart contract execution. Web3Agent incorporates real-time inspection, error handling, and interaction transparency across its operation log, and flow visualization components. We evaluate the system and perform ablation study with customized dataset in a simulated environment, demonstrating its feasibility in orchestrating complex Web3 tasks and highlighting implications for agent-based abstraction in decentralized systems.
<p>This Zenodo record contains the verified smart contract source code for the TWLGF token project.</p> <p>The code has been published and verified on BscScan as an Exact Match. <br>This record provides an EU OpenAIRE DOI and long-term archive reference for the official contract source.</p> <p>Network: BNB Smart Chain (BSC)<br>Compiler: solc v0.5.16+commit.9c3226ce<br>Optimization: 200 runs<br>License: MIT</p> <p>Verified contract address:<br>https://bscscan.com/address/0xA45D0f9337eF5539d77c41e968137C391d1d7704#code</p> <p>This record complements other publicly archived TWLGF documentation:<br>• Internet Archive: https://archive.org/details/twlgf-bsc-scan-verified-source-2025-11-12<br>• GitHub repository: https://github.com/twlgfofficial/twlgf_logo<br>• Google Drive timestamped PDF (2025-11-12)<br>• Whitepaper DOI: https://zenodo.org/records/17516133</p>
This paper introduces a structured approach to improving decision making in Decentralized Autonomous Organizations (DAO) through the integration of the Question-Option-Criteria (QOC) model and AI agents. We outline a stepwise governance framework that evolves from human led evaluations to fully autonomous, AI-driven processes. By decomposing decisions into weighted, criterion based evaluations, the QOC model enhances transparency, fairness, and explainability in DAO voting. We demonstrate how large language models (LLMs) and stakeholder aligned AI agents can support or automate evaluations, while statistical safeguards help detect manipulation. The proposed framework lays the foundation for scalable and trustworthy governance in the Web3 ecosystem.
AI's merger with Web3 tech is changing the game leading to open, see-through, and somewhat self-running systems. This research explores the AI-agent paradigm in Web3 considering spread-out networks, blockchain rules, smart deals, and partial self-rule as new concepts. The main focus of this paper is to redefine the integration of artificial intelligence (AI) with Web3 technologies and create a semi-autonomous architecture that transcends decentralized and centralized approaches. While the majority of literature surveys AI agents that function on blockchain and decentralized protocols, our study presents a layered model that exploits off-chain AI inference with on-chain consensus mechanisms, (DID) management, and governance. We explain the main components, such as shared record-keeping, distributed ID management, reward systems, and agreement methods, that allow AI agents to function efficiently in the absence of a central boss. Moreover, the research analyzes significant issues such as scalability, safety, data privacy, and interoperability. It offers a number of improvements in off-chain AI-based agents for decentralized environments. Results demonstrate that hybrid on-chain/off-chain AI clusters can reduce inference costs and increase transaction throughput while preserving decentralization and data privacy. The team-up of AI and Web3 opens doors to new uses like spread-out money systems (DeFi), self-running groups (DAOs), and marketplaces without middlemen creating tough, clear, and user-focused digital worlds.
Lorena Baigorria, Ana Gabriela Garis, Daniel Eduardo Riesco
Blockchain es la base tecnológica de una nueva forma de realiz-ar transacciones de manera segura en una red descentralizada. Dicha tecnología permite registrar la validez y el origen de los datos, y realizar transacciones de manera digital, compartida, inalterable y sin la intervención de intermediarios. Frecuentemente, las transacciones requieren de lógica automa-tizada. En estos casos, se vuelve necesaria la definición de con-tratos inteligentes, programas de computación almacenados en Blockchain que se ejecutan automáticamente cuando se cumplen condiciones predeterminadas. Los errores en contra-tos inteligentes pueden tener graves consecuencias, especial-mente en ámbitos como finanzas descentralizadas. Una clara definición de las condiciones es esencial; sin embargo, éstas son generalmente descriptas en lenguaje natural por las partes involucradas, lo que conlleva a la ambigüedad de interpretación por parte de los programadores del contrato. Por otro lado, los errores en la programación también pueden derivar a que el contrato no se ejecute como se esperaba. La calidad del contra-to inteligente podría ser mejorada si las condiciones fueran especificadas en UML con OCL, y luego transformadas al len-guaje de modelado Alloy para llevar a cabo la verificación y validación formal a través del método Model Checking. En este artículo, se describe una línea de investigación que propone un modelo para la especificación de contratos inteligentes en UML y OCL, complementado con una transformación automática a Alloy para su verificación y validación. Dicho modelo contribuye a realizar una auditoría más rigurosa de contratos inteligentes antes de despliegue en Blockchain.
To address the issues of inefficiency and high debugging costs in smart contract development, this paper proposes a multi-agent collaborative solution based on large language models (LLMs). In this approach, agents collaborate to accomplish tasks such as smart contract generation, auditing, and repair. First, an efficient fine-tuning method is employed to optimize the code generation model. Second, retrieval-augmented generation technology is utilized to provide vulnerability context knowledge for code repair. Finally, a feedback-driven multi-agent collaboration framework is designed to enhance code generation quality through agent-to-agent feedback. During the code repair phase, semantic prompts derived from control flow graphs guide the LLM in contract repair. Experimental results demonstrate that the multi-agent architecture significantly enhances the capabilities of smart contract generation and auditing. By leveraging knowledge of vulnerabilities and external tools, the approach improves the LLM's understanding of smart contract structures. Compared to a single LLM approach, the generated code shows significant improvements in CodeBLEU scores, compilation pass rates, and vulnerability rates.
The dominant paradigm of the modern internet, built on graphical user interfaces (GUIs) and discrete web applications, forces users into a fragmented and manual process of interaction, fundamentally limiting the complexity of tasks that can be automated. This paper argues for a new architectural paradigm: an AI-powered network of autonomous digital agents that replaces manual navigation with goal-oriented, natural language-based service procurement. To establish the necessity and viability of this approach, a critical literature review of dominant service-oriented architectures is conducted. The analysis reveals that Microservice Architecture (MSA), despite its advantages, suffers from inherent complexities in communication, discovery, and data management that undermine true service autonomy. A review of its predecessors finds that Service-Oriented Architecture (SOA) was hindered by centralized bottlenecks, while the Semantic Web’s vision of a machine-readable web failed due to the rigidity and complexity of its formal, logic-based approach. Furthermore, modern Web3 architectures, while offering decentralization, are shown to have severe limitations in scalability, cost, and flexibility that make them unsuitable for dynamic agent collaboration. The paper concludes that these existing paradigms contain fundamental gaps and posits that a Multi-Agent System (MAS) architecture provides a more robust and appropriate foundation for building a truly autonomous, post-GUI digital ecosystem.
Javier Godoy, Margarita Capretto, Martín Ceresa, Juan Pablo Galeotti · 7 authors
Smart contracts manage valuable assets, and their immutability hinders bug fixing. Therefore, pre-deployment verification and validation are critical. In fact, auditing has become mandatory in the pipeline of smart contract development. Auditors usually combine manual inspection with automated tools in their auditing work, looking for issues that may be domain dependent (i.e., pertaining to the correct implementation of requirements-which are often informal, partial, and implicit) or independent (e.g., reentrancy, overflow, etc.), To identify domain dependent issues, it is important to understand the non-trivial behavior of the implementation over sequences of calls made by callees playing different roles in the contract. In this paper, we propose a novel approach that combines predicate abstraction with modal transition systems to build abstractions that can help auditors in the smart contract validation process. The required inputs are a set of predicates provided as code and, optionally, constraints over smart contract function parameters. The output is a modal transition system that captures the contract's behavior. We report on a prototype that builds modal abstractions and an evaluation on two established benchmarks where we identified four previously unreported issues.
Aniket P. Kakde, Karan M. Bhoyar, Muhammad Aiman Shad, Prof. Sudesh A. Bachwani
Autonomous agents powered by Large Language Models (LLMs) require reliable and standardized frameworks to connect tools, exchange contextual information, and synchronize tasks across diverse systems. Despite growing interest in such agents, current integration with external tools remains disjointed. Developers often have to manually create interfaces, handle authentication protocols, and navigate incompatible function-calling standards across platforms. To overcome these limitations and promote the evolution of agentic AI, it is critical to establish standardized communication protocols that ensure interoperability—enabling agents and systems to seamlessly discover each other’s capabilities, share data, and coordinate operations. This paper explores a structured overview of emerging communication standards for agents, focusing on the Model Context Protocol (MCP), Agent Communication Protocol (ACP), Agent-to-Agent Protocol (A2A), and Agent Network Protocol (ANP). MCP utilizes a JSON-RPC based client-server architecture to enable secure execution of tools and well-typed data transfer. ACP introduces a REST-compliant message structure with support for asynchronous streaming and multipart formats, facilitating rich, multimodal agent outputs.A2A enables agents to delegate tasks peer-to-peer using capability-rich Agent Cards, enabling scalable and distributed workflows across organizations. ANP facilitates agent discovery and secure collaboration in open networks, leveraging decentralized identifiers (DIDs) and semantic graphs based on JSON-LD.
Multi-Agent AI Systems (MAS) rely on the cooperative actions of autonomous agents to meet difficult and rapidly changing issues in analysis and business strategy. In contrast to single-agent models, MAS includes different agents that team up, change as needed and function in real time. Thanks to its decentralized and modular design, businesses can scale their activities, maintain good stability and flex their operations as market situations change. With the help of advanced AI like Generative AI, MAS can examine huge datasets, perform market simulations and support smart decisions from leaders. Such algorithms are applied to everything from setting creative prices to improving supply chains, assessing risks and detecting fraud in the financial industry. The use of MAS makes it possible for tasks to be split and completed by multiple processors, which helps reduce workflow trouble spots. Additionally, its ability to respond to uncertainty and make quick, real-world decisions makes MAS a vital instrument for industries needing both agility and innovation. With MAS, organizations become stronger competitors by streamlining their work processes, encouraging innovation and solving problems on many scales. The future success of MAS comes from its power to change how businesses run smoothly by working with present technology and developing together with the company's needs.
Smart contracts underpin contemporary decentralized systems, yet their immutability and perpetual execution amplify the consequences of latent defects. Despite progress in manual audits, static analysis, fuzzing, and formal verification, auditors face a widening gap between the scalability and desired assurance due to limited automation capability. Recent large language models (LLMs) with tool-use capabilities promise greater automation, but monolithic single-agent auditors struggle with coverage, robustness, and reproducibility. Motivated by addressing these issues, we propose Multi-Agent Auditing (MAA), a framework that coordinates a team of tool-grounded agents through a constrained protocol that privileges verifiable artifacts. Besides, we mechanize an LLM-assisted orchestration mechanism and a shared knowledge base to coordinate a set of agents specialized in sophisticated testing approaches to produce budget-aware and evidence-centric audit results. Furthermore, we present the experiment results showing that MAA outperforms singleLLM auditors and provide empirical insights into LLM-backend selection.
Decentralized Autonomous Organizations (DAOs) are a class of Decentralized Applications (DApps) using smart contracts to facilitate governance processes. The design of DAOs is affected by additional complexity compared to other DApps due to the need to specify organizational roles, permissions, and control relations early in the early development stages of the system. In addition, DAOs face scalability challenges. While existing Model-Driven Development (MDD) tools support general smart contract and DApp design, they lack constructs tailored to the organizational and governance features unique to DAOs. To bridge this gap, we develop a code generation approach for DAO-ML, a visual modeling language for DAO design. The translator we implement for this method generates smart contracts of DAOs with suitably configured roles and permissions from visual models. The generated smart contracts particularly optimize the representation of roles and permissions to improve the system scalability and handle complex governance structures, necessary to increase the utility of DAO systems. The approach is evaluated in the context of an in vivo case study on the development of a DAO for the disintermediated management of local tourism. This work advances MDD for decentralized systems by bridging high-level governance modeling with executable, gas-efficient smart contract code generation.
Quentin Botha, Laurent Bindschaedler, Christoph Siebenbrunner
Decentralized mathematics prediction markets promise new forms of collaboration and incentive alignment, but traditional requirements engineering methods fail to address the unique governance, incentive, and security challenges of such Web3 systems. This paper demonstrates how they can be addressed through a requirements-driven design of a decentralized prediction market for mathematical conjectures, and proposes concrete enhancements to existing frameworks. Our work delivers actionable guidelines for engineering secure, incentive-aligned decentralized platforms, and sets a new standard for early-stage RE in the Web3 era.
The objective of this study is to define a compliance-first, conceptually generalisable architecture for a multi-agent artificial intelligence platform integrated with distributed ledger technology, designed to be domain-, deployment-, and vendor-agnostic. It addresses a persistent shortcoming in current AI deployments, where compliance is often treated as a secondary concern, applied retroactively through prompt engineering rather than embedded within the foundational design. The proposed model encodes regulatory, governance, and ESG requirements into an objective-under-constraints framework, ensuring that all specialised agents operate within legally admissible and verifiably auditable parameters prior to any domain-specific implementation. A DAG-based verification layer is incorporated to enable scalable, low-latency, and cost-efficient operation while preserving evidentiary integrity. The analysis evaluates the feasibility of this conceptual model to support sustainable, rapid-deployment vertical applications without inducing vendor lock-in, preserving operational neutrality, and ensuring environmental accountability. The findings suggest that integrating compliance, ESG metrics, and agent specialisation at the architectural level provides a transferable foundation for cross-domain AI-DLT infrastructures.
The US market recently moved to a T+1 settlement cycle, which is quite a big step forward, but the real goal here is achieving real time settlement - basically T + 0 or instant settlement - with the ultimate goal of achieving real-time (T+0) settlement, which could significantly enhance the efficiency and stability of U.S. financial markets. This paper proposes a complete solution for instant trade settlement in the US market. Our approach involves using technologies like Distributed Ledger Technology (DLT) and Blockchain and combining them with intelligent automation using Agentic Artificial Intelligence (AI) and Large Language Models (LLMs) for their analytical power. The system being detailed here is built around a private, permissioned blockchain setup. This approach gives us immutable record keeping and atomic settlement capabilities through smart contracts. What this means is that tokenized securities and cash can be transferred instantaneously. When it comes to Agentic AI, this technology is being looked at to handle various tasks autonomously. Tasks like pre trade validation, managing real time liquidity, optimizing collateral, and dealing with post trade exceptions. These capabilities collectively aim to reduce both operational and counterparty risks. As for LLMs, these will work as intelligent interfaces for monitoring and reporting. They will handle regulatory compliance automation and improve market insights using natural language processing. This whole integrated framework being proposed should help drastically cut down counterparty risk, boost liquidity, reduce operational costs, and make things more transparent and auditable. This paper also tackles some key challenges - regulatory alignment, interoperability issues, scalability concerns, and cybersecurity aspects. This provides a realistic roadmap for where U.S. financial market infrastructure might head in the future. This work introduces a comprehensive framework that brings together permissioned distributed ledger technology, autonomous AI agents, and large language models to streamline regulatory processes. The result is a real-time (T+0) settlement system specifically designed to meet the demands of institutional trading in the U.S. financial markets.
This research article presents a novel architecture to empower multi-agent economies by addressing two critical limitations of the emerging Agent2Agent (A2A) communication protocol: decentralized agent discoverability and agent-to-agent micropayments. By integrating distributed ledger technology (DLT), this architecture enables tamper-proof, on-chain publishing of AgentCards as smart contracts, providing secure and verifiable agent identities. The architecture further extends A2A with the x402 open standard, facilitating blockchain-agnostic, HTTP-based micropayments via the HTTP 402 status code. This enables autonomous agents to seamlessly discover, authenticate, and compensate each other across organizational boundaries. This work further presents a comprehensive technical implementation and evaluation, demonstrating the feasibility of DLT-based agent discovery and micropayments. The proposed approach lays the groundwork for secure, scalable, and economically viable multi-agent ecosystems, advancing the field of agentic AI toward trusted, autonomous economic interactions.
In increasingly dynamic organizational and computational environments, workflows controlled by multi-agent systems face often face context changes, blurred task borders, and unexpected behaviors. This research proposes a new model for coping with decision drift in adaptive workflows by modeling latent boundary negotiation. By using concepts from complex systems theory, cognitive modeling, and distributed artificial intelligence, we define agents as autonomous systems which are able to perceive, report, and re-negotiate internal role expectations through localized negotiation protocols. The designed model merges drift detection with multi-agent coordination techniques and correlates them to simulated workflow processes typical for high risk, time constrained settings like crisis management, distributed logistics, and agile development teams. Simulation results show that latent negotiation improves task coherence, decreases agent interference, and stabilizes the performance of workflows under ambiguous and changing boundary conditions much more than negotiation less systems. The benefits of negotiation-aware multi-agent systems in supporting sustained alignment of decisions and flexibility in the system are highlighted through comparison with static and rule-based systems. The results outline latent boundary negotiation as a key feature towards the implementation of intelligent self-organizing work systems in decentralized multi agent ecosystems.
Theorem proving serves as a major testbed for evaluating complex reasoning abilities in large language models (LLMs). However, traditional automated theorem proving (ATP) approaches rely heavily on formal proof systems that poorly align with LLMs' strength derived from informal, natural language knowledge acquired during pre-training. In this work, we propose DeepTheorem, a comprehensive informal theorem-proving framework exploiting natural language to enhance LLM mathematical reasoning. DeepTheorem includes a large-scale benchmark dataset consisting of 121K high-quality IMO-level informal theorems and proofs spanning diverse mathematical domains, rigorously annotated for correctness, difficulty, and topic categories, accompanied by systematically constructed verifiable theorem variants. We devise a novel reinforcement learning strategy (RL-Zero) explicitly tailored to informal theorem proving, leveraging the verified theorem variants to incentivize robust mathematical inference. Additionally, we propose comprehensive outcome and process evaluation metrics examining proof correctness and the quality of reasoning steps. Extensive experimental analyses demonstrate DeepTheorem significantly improves LLM theorem-proving performance compared to existing datasets and supervised fine-tuning protocols, achieving state-of-the-art accuracy and reasoning quality. Our findings highlight DeepTheorem's potential to fundamentally advance automated informal theorem proving and mathematical exploration.
Luigi Bellomarini, Marco Favorito, Giuseppe Galano
We present BitML2MCMAS, a formal verification tool for analyzing Bitcoin smart contracts, when specified in BitML, through ATL model checking using the MCMAS model checker. We developed a translation procedure from a BitML contract to an MCMAS model that simulates the BitML semantics, allowing for strategic reasoning on BitML smart contracts. We tested our tool over several case studies, showing that we can verify smart contract specifications that capture interesting multi-agent strategic interactions.
This paper addresses the challenge of creating smart contracts for applications represented using Business Process Management and Notation (BPMN) models. In our prior work we presented a methodology that automates the generation of smart contracts from BPMN models. This approach abstracts the BPMN flow control, making it independent of the underlying blockchain infrastructure, with only the BPMN task elements requiring coding. In subsequent research, we enhanced our approach by adding support for nested transactions and enabling a smart contract repair and/or upgrade. To empower Business Analysts (BAs) to generate smart contracts without relying on software developers, we tackled the challenge of generating smart contracts from BPMN models without assistance of a software developer. We exploit the Decision Model and Notation (DMN) standard to represent the decisions and the business logic of the BPMN task elements and amended our methodology for transformation of BPMN models into smart contracts to support also the generation script to represent the business logic represented by the DMN models. To support such transformation, we describe how the BA documents, using the BPMN elements, the flow of information along with the flow of execution. Thus, if the BA is successful in representing the blockchain application requirements using BPMN and DMN models, our methodology and the tool, called TABS, that we developed as a proof of concept, is used to generate the smart contracts directly from those models without developer assistance.