Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

AgentDAO: Synthesis of Proposal Transactions Via Abstract DAO Semantics

Lin Ao, Han Liu, Huafeng Zhang

TL;DR

AgentDAO addresses the barrier to DAO governance proposal participation by translating natural language into executable transactions through a three-layer LLM-based architecture. It introduces DAOLang, a high-level DSL with symbolic evaluation to produce transaction payloads, and a Label-Centric Retrieval (LCR) mechanism that leverages semantic labels to select relevant DAOLang samples for synthesis. In extensive preliminary evaluation on CompoundV3, AgentDAO achieves up to $89.68 ext{\%}$ success with $k=4$, demonstrating strong performance for composite and cross-chain proposals while revealing trade-offs in gas efficiency and handling of ambiguous commands. The work highlights the practical potential of structured language representations and multi-agent prompting to democratize governance actions across multi-chain ecosystems, while outlining current limitations and directions for extension to broader protocols.

Abstract

While the trend of decentralized governance is obvious (cryptocurrencies and blockchains are widely adopted by multiple sovereign countries), initiating governance proposals within Decentralized Autonomous Organizations (DAOs) is still challenging, i.e., it requires providing a low-level transaction payload, therefore posing significant barriers to broad community participation. To address these challenges, we propose a multi-agent system powered by Large Language Models with a novel Label-Centric Retrieval algorithm to automate the translation from natural language inputs into executable proposal transactions. The system incorporates DAOLang, a Domain-Specific Language to simplify the specification of various governance proposals. The key optimization achieved by DAOLang is a semantic-aware abstraction of user input that reliably secures proposal generation with a low level of token demand. A preliminary evaluation on real-world applications reflects the potential of DAOLang in terms of generating complicated types of proposals with existing foundation models, e.g. GPT-4o.

AgentDAO: Synthesis of Proposal Transactions Via Abstract DAO Semantics

TL;DR

AgentDAO addresses the barrier to DAO governance proposal participation by translating natural language into executable transactions through a three-layer LLM-based architecture. It introduces DAOLang, a high-level DSL with symbolic evaluation to produce transaction payloads, and a Label-Centric Retrieval (LCR) mechanism that leverages semantic labels to select relevant DAOLang samples for synthesis. In extensive preliminary evaluation on CompoundV3, AgentDAO achieves up to success with , demonstrating strong performance for composite and cross-chain proposals while revealing trade-offs in gas efficiency and handling of ambiguous commands. The work highlights the practical potential of structured language representations and multi-agent prompting to democratize governance actions across multi-chain ecosystems, while outlining current limitations and directions for extension to broader protocols.

Abstract

While the trend of decentralized governance is obvious (cryptocurrencies and blockchains are widely adopted by multiple sovereign countries), initiating governance proposals within Decentralized Autonomous Organizations (DAOs) is still challenging, i.e., it requires providing a low-level transaction payload, therefore posing significant barriers to broad community participation. To address these challenges, we propose a multi-agent system powered by Large Language Models with a novel Label-Centric Retrieval algorithm to automate the translation from natural language inputs into executable proposal transactions. The system incorporates DAOLang, a Domain-Specific Language to simplify the specification of various governance proposals. The key optimization achieved by DAOLang is a semantic-aware abstraction of user input that reliably secures proposal generation with a low level of token demand. A preliminary evaluation on real-world applications reflects the potential of DAOLang in terms of generating complicated types of proposals with existing foundation models, e.g. GPT-4o.

Paper Structure

This paper contains 28 sections, 7 equations, 6 figures, 1 table, 1 algorithm.

Table of Contents

  1. Introduction
  2. The Decentralized Autonomous Governance Language (DAOLang)
  3. Syntax
  4. Symbolic Evaluation
  5. System Design
  6. Relation Extractor
  7. Program Synthesizer
  8. Samples Construction.
  9. Normalization.
  10. Assigning Labels.
  11. Embedding.
  12. The LCR algorithm.
  13. Prompt Construction.
  14. Semantic Interpreter
  15. Evaluation
  16. ...and 13 more sections

Figures (6)

  • Figure 1: Initate a proposal with the payload on Compound
  • Figure 2: A representative set of the syntax of DAOLang
  • Figure 3: A representative set of the symbolic evaluation of the DAOLang program
  • Figure 4: Illustration of the architecture of AgentDAO. User utterance is translated into the DAOLang program and is reformulated into a transaction payload.
  • Figure 5: An illustrative example of LCR. The LCR algorithm first ranks samples according to their vector space and then ranks them in descending order based on the marginal contribution matched labels
  • ...and 1 more figures

Theorems & Definitions (4)

  • Example 1
  • Example 2
  • Example 3
  • Example 4