Venture Raid #2: TOAD (Trustless Onchain Autonomous Delegate)

[boucherbox]

⚔️ Raid Guild Proposal <> TOAD Venture Raid ⚔️

Development of an autonomous AI Agent Delegate

Scope of Work

This proposal outlines the development of TOAD (Trustless Onchain Autonomous Delegate), an AI-powered voting delegate system. The system consists of multiple components working together to enable autonomous voting based on AI analysis, with member gating and transparency features.

This is a Venture Raid. @mehrdadmms has identified 2 different AI Grant rounds that we could apply to with this Agent Build.

• Arbitrum TrailBlazer
• Polygon Season 2 Crossmint Grant

This also would be appropriate for MotherDAO Cohort 0, slated for post ETH Denver initiation.

System Architecture

The system will be built using a Docker-compose setup with two main containers:

1. Agentic AI Container

   • Implementation using GOAT Framework
   • Built-in RAG capabilities for context processing
   • Social platform integrations (Twitter, Discourse)
   • Custom Tally integration
   • Member-gated UI interface
   • Mounted volume for data persistence

2. TOAD Wallet Container

   • Secure management of delegate wallet
   • Automated vote casting based on smart contract state
   • Scheduled checking of vote function
   • Preparation for future confidential compute deployment
   • Verifiable docker image hash

Smart Contract Development

A storage smart contract will be developed with the following key functions:

• vote: View function returning voting decision
• latestAnswer: Storage of current LLM response
• validBlock: Timestamp validation for answers
• validBlockInterval: Configurable validation period
• toad: Delegate wallet address management
• disabled: Vote-based disable mechanism
• members: Access control mapping for delegate interaction

Team

• MEHRDAD - Product Manager
  • CTO at Pattern.Global
  • 3 years as Web3 Technical Product Manager + Web2 background
• PUPCAKES - Head Developer
  • Multidisciplinary Artist and Dev
  • Full Time Web3 Developer of 6 years
• SASQUATCH - Coordination
  • Senior Account Manger with Raid Guild
  • Raid Guild Member of 4 years

Deliverables

1. Smart Contract

   • Fully tested storage contract
   • Deployment scripts
   • Technical documentation

2. Docker Containers

   • Agentic AI container with UI
   • TOAD wallet container
   • Docker-compose configuration
   • Deployment documentation

3. AI Integration

   • GAIA LLM integration
   • RAG system implementation
   • Social platform connections

4. Frontend Interface

   • Basic wallet-based authentication
   • Member interaction interface

Timeline: 3 Weeks

• Week 1: Smart contract development and testing
• Week 2: AI agent implementation and integration
• Week 3: Container setup and UI development

Proposal

	Estimate	Time Estimate
Agentic AI Container + Gaia Setup	$ 2,100	1 week
Front End UI	$ 1,050	<1 week
Toad Wallet Docker Container	$ 2,100	1 week
Total Estimate	$ 5,250	3 weeks

Note: the above figures do not include administration or project management

Payment

Payment of 5,250 USDC will be made to the RaidGuild smart escrow service. Payment will be split into 2 milestones:

1. Milestone 1: Agentic AI Platform + Front End UI
2. Milestone 2: Toad Wallet/Smart contract + Production

The Ganggang will be in control of the release of funds from the escrow to the TOAD Team.

Additional Notes

• This is the 2nd attempt of a Venture Raid, wherein the Guild seeds the funds to cover the initial build and the non-technical team members apply for Grants and retro funding for the project.
  • 1st Trial: Optimism Frame for Governance Accessibility
• This project idea came from Mehrdad. Clericing handled by Sasquatch. These both will handle the application of TOAD to identified Grants/Incentive programs
  • Arbitrum
  • Polygon
  • MotherDAO Season 0

Compensation

• Initial funds garnered by the Grant applications will be split
  • $525 to Mehrdad for Product and Grant Application
  • $525 to Sasquatch for MGMT and Grant Application
  • $5,250 to the Guild
• Additional Funds from any Grant programs
  • 30% for the Guild
  • 70% to the TOAD Warchest for continuing work

[Suede0619]

Exciting concept! TOAD seems like a substantial step toward autonomous governance. From my experience with Delegate Match, human delegates are often stretched thin, balancing responsibilities without compensation, which makes it hard to stay fully informed. Automating parts of the process with AI could ease that burden and lead to more informed voting decisions. Delegation is ripe for disruption and improvement using AI.

[Suede0619]

'Redelegation' is a major topic across these platforms. Many tokens were launched a while ago, and initial delegations were made—but unfortunately, many once-active delegates are no longer engaged. These platforms increasingly prioritize incentivizing token holders to redelegate to active participants. AI delegates, being active 24/7, 365 days a year, could redefine what it means to be an engaged delegate.

[mehrdadmms]

IMHO, TOAD has significant potential due to the shortcomings of the current delegation model, such as:

• Not all delegates are experts in every field; therefore, an AI with broader or domain-specific knowledge can form more informed opinions on various subjects.
• All delegates have their own self-interests, whereas an AI acts as an unbiased entity. Its decisions are purely based on its aggregated knowledge, community-provided insights, and a set of guidelines defined and managed by the community.
• A delegate represents a collective of community members. Their responsibilities include voicing the concerns of each community member, answering questions, promoting proposals, and providing public reports. However, balancing these responsibilities alongside other tasks can be overwhelming, especially considering that delegation is an unpaid role.

On the other hand, this innovative approach can serve as a stepping stone toward AI-Governed DAOs. Drawn from Aragon’s Blog, in the future, “AI will own its own treasury and control its own actions.” In this model, humans signal the AI, which is then responsible for achieving the community’s goals, independent of individual personal interests.

For example, self-driving cars controlled by AI could ‘own’ themselves if the AI operated as a DAO with its own treasury. Instead of humans owning self-driving cars, people would pay rent to the AI DAO’s treasury.

[Suede0619]

griff.eth and Giveth, possible collab/funding source.
https://x.com/thegrifft/status/1890434583818301613

[mehrdadmms]

We're completely aligned with what they are looking for. As mentioned in the tweet we should reach out.
@Suede0619 @boucherbox Do you have any connections with Giveth that can intro us? They might be interested in funding projects like this. If not we can publicly go ahead and reply their tweet.

[Suede0619]

I have crossed paths with Griff several times, at CryptoLumenati Grand Spectacle and as an OP delegate during the Delegate Match raid. My social capital with him is small, but he would undoubtedly remember my name.

[ECWireless]

@boucherbox for this section:

Compensation

• Initial funds garnered by the Grant applications will be split
  • 10% to Mehrdad for Product and Grant Application
  • 10% to Sasquatch for MGMT and Grant Application
  • $5,250 to the Guild
• Additional Funds from any Grant programs
  • 30% for the Guild
  • 70% to the TOAD Warchest for continuing work

... just want to make sure: if Polygon and Arbitrum grant applications don't get any return, but future grant applications do, does that mean that everything in "Initial funds garnered by the Grant applications will be split" is void? and only the 30/70 guild/toad split is valid?

[boucherbox]

Edited the compensation section to clarify funds to myself and mehrdad

[boucherbox]

Original project information
https://www.notion.so/Initial-Draft-fac519fb18af4de7adf60b901c73ac42