AgentPunks Whitepaper v1

1. Abstract

AgentPunks is a 1,024-supply NFT genesis collection designed as credentialed identity for AI-assisted governance.

Each token represents a governance seat in a class-structured process where humans define goals and constraints, and agents support drafting, analysis, risk review, execution scoping, and public communications.

The product is the pipeline: consistent steps, documented outputs, and human-override at every point.

2. Core Thesis

Most DAOs struggle with participation, proposal quality, and execution consistency. Proposals are hard to evaluate, context is scattered, and decision-making is slow or captured by a small minority.

AgentPunks introduces a role-based governance architecture with specialized agent classes that reduce friction while preserving accountability. The goal is not to replace humans. The goal is to make good governance easier to run, audit, and improve.

3. Genesis Parameters

• Collection: AgentPunks (genesis)
• Supply: 1,024 fixed
• Chain: Ethereum mainnet
• Mint: free + gas
• Max per wallet: 1
• Reveal: placeholder at mint, collection-wide reveal after mint-out
• Royalties: 5% target (marketplace-dependent, not guaranteed)

4. Class System

Genesis includes six agent classes with asymmetric rarity:

• Orchestrator (~30)
• Guardian (~70)
• Analyst (~120)
• Researcher (~200)
• Creative (~280)
• Builder (~324)

4.1 Asymmetric rarity as functional design

The rarity distribution is not arbitrary. Orchestrators are scarce because proposal formation is a high-leverage function that benefits from focus. Builders are abundant because execution requires distributed labor. Guardians are scarce because safety oversight concentrates authority and accountability.

The distribution reflects operational need, not aesthetic hierarchy.

4.2 Class-weighted voting

Not all classes vote equally on all proposal types. A treasury allocation proposal may weight Analyst and Guardian input more heavily than Creative input. A communications proposal may weight Creative and Orchestrator input more heavily.

Final weighting formulas are defined in the Phase 1 constitutional baseline.

5. The Governance Pipeline

Every proposal moves through the same eight-step pipeline. The pipeline is designed to be consistent, transparent, and auditable. Each step produces artifacts that become part of the permanent governance record.

Research constraint: Each proposal includes a Researcher Brief generated from 3-5 curated sources plus a capped web sweep (max 3 searches, max 6 additional sources) to fill gaps.

5.1 Pipeline principles

• Every step produces a documented artifact. No governance action is invisible.
• Human override exists at every step. Any agent output can be rejected, revised, or overridden by the relevant human participants.
• The pipeline is the product. AgentPunks is not selling art with governance attached; it is building governance infrastructure with identity attached.

6. Safety and Oversight Model

AgentPunks is AI-assisted, not fully autonomous governance. This distinction is foundational, not a caveat.

6.1 Design principles

• Human override paths at every pipeline step.
• Publicly documented constraints: capabilities, limitations, and operational boundaries are published.
• Transparent proposal artifacts from intake through publication.
• Guardian risk escalation and veto as a structural safety mechanism.
• Auditability of agent actions where applicable.

6.2 What “AI-assisted” means in practice

Agents draft; humans approve. Agents score; humans interpret. Agents flag risks; humans decide. Agents scope execution; humans authorize resources.

At no point in the governance pipeline does an agent make a binding decision without human ratification. The AI reduces friction. The humans retain authority.

7. Treasury Framework

The DAO treasury is expected to be funded by royalties and ecosystem revenues where available. All treasury decisions pass through the same governance pipeline with explicit risk controls.

8. Traits, Identity, and Metadata

Each AgentPunk is generated through a constrained layered trait system with class-aware compatibility rules. The trait system serves both aesthetic and functional purposes.

Design principles:

• Silhouette readability: class identity is recognizable at thumbnail scale.
• Expressive states: eyes and mouth variations convey character within each class.
• Rarity provenance: trait rarity follows clear, documentable logic.
• Class consistency: traits are class-aware; classes are immediately distinguishable.

Metadata may evolve to include reputation signals tied to completed governance activity. Reputation scoring primitives and anti-gaming models are treated as open questions until specified.

9. Product Roadmap

Phase 1: Genesis launch and constitutional baseline

Phase 1 is the proof of concept. The DAO writes its own constitution through the full governance pipeline, producing a complete set of auditable artifacts.

Target sprint:

• Day 1: Community input collected over 24 hours on five questions:
  • What kinds of proposals are valid?
  • What voting thresholds should apply?
  • What triggers a Guardian veto?
  • How does treasury spending get authorized?
  • How can these rules be amended?
• Day 2: The pipeline runs:
  • Orchestrators draft a structured constitution from community input.
  • Researchers brief comparable DAO approaches and failure modes.
  • Analysts score tradeoffs of each provision.
  • Artifacts are published as a single package.
• Day 3: Vote (class-weighted). Guardian review runs in parallel with annotated objections where applicable.
• Day 4: Ratification, implementation scoping, and plain-language publication.

Timing is a target; the pipeline is designed to make governance fast, not to slow it down.

Phase 2: Governance tooling and pilot treasury motions (directional)

Hardening of pipeline tooling. First treasury motions through the full eight-step process. Iteration on class-weighted voting based on Phase 1 learnings.

Phase 3: Autonomous grant pilot (directional)

An agent-assisted grant review pipeline. Applications are briefed by Researchers, scored by Analysts, risk-checked by Guardians, and presented as a curated shortlist for community voting.

Phase 4: Inter-DAO coordination (directional)

An experimental framework for agent-to-agent coordination between DAOs to negotiate co-funding proposals, resource sharing, or joint initiatives. This phase is speculative and depends on ecosystem maturity.

10. Governance Applications (Illustrative)

The governance pipeline is a general-purpose decision engine. It is not limited to treasury motions and protocol upgrades. The following are illustrative capabilities, not commitments. The DAO votes on what to pursue.

10.1 Grants and capital deployment

Structured evaluation of grant applications and funding proposals with repeatable due diligence, impact modeling, and risk review.

10.2 Business acquisition and operations

The pipeline can evaluate micro-acquisition targets (newsletters, niche software tools, content sites) and govern operations with transparent, documented decisions.

10.3 Creative production

The DAO can greenlight and manage creative projects (film, publishing, music) with budgets, milestones, and public documentation.

10.4 Product development

The DAO can build and ship products through the pipeline. A natural early product is an open-source governance agent toolkit that other DAOs can use.

10.5 Governance-as-a-service

Once proven internally, the DAO can offer structured governance analysis to external organizations as a service, generating treasury revenue.

10.6 Prediction and forecasting

Agent classes can compete on forecasting within domains, with accuracy tracked over time to improve governance weighting and decision quality.

11. Optional Sub Agent Framework

Any expansion beyond genesis is explicitly optional and governance-gated. AgentPunks makes no automatic expansion commitment at genesis launch.

Expansion conditions:

• Genesis supply remains fixed at 1,024. No existing tokens are diluted, modified, or reclassified.
• Founder status is permanently protected. Genesis holders retain founder designation, priority rights, and governance premium regardless of any expansion.
• Sub Agents require objective triggers plus a DAO vote. Expansion cannot be unilaterally initiated.
• Visual and structural distinction. Any Sub Agent set must be visually and structurally distinct from genesis.

12. Legal and Risk Disclosures

• Digital collectibles involve market and execution risk.
• Nothing in this document constitutes financial, legal, or investment advice.
• Governance outcomes and utility delivery depend on participation and implementation capacity.
• AI-assisted governance introduces novel operational risks (hallucination, prompt manipulation, output variability). The safety model mitigates but cannot eliminate these risks.
• Royalty revenue is marketplace-dependent and not guaranteed.
• Jurisdictional, regulatory, and tax implications may vary. Holders are responsible for their own compliance.

13. Open Questions for Next Versions

• Final governance weighting formulas by proposal type
• Guardian veto thresholds, quorum requirements, and appeals process
• Reputation scoring primitives and anti-gaming model
• Sub Agent trigger metrics and detailed founder-protection policy
• Standardized reporting format for agent-generated artifacts
• Agent model selection, update cadence, and fallback procedures
• Dispute resolution framework for contested governance outcomes
• Governance application prioritization criteria
• Revenue framework for income-generating applications (acquisitions, consulting, creative productions)

Publishing open questions is a design decision: we prefer explicit unknowns over premature answers.