Lil Nouns DAO

Lil Nouns already has an AI Agent live inside Farcaster DMs. It works, but because it has no structured DAO data to draw from, replies often hallucinate.

This proposal is not just about fixing one bot. It is about building AI-native data infrastructure: governance, community, cultural, and developer data structured into pipelines, stored both off-chain for scale and on-chain for provenance, indexed for retrieval, and exposed through a Model Context Protocol (MCP) server. This eliminates hallucinations and makes Lil Nouns data portable across any AI workflow.

The infra will also integrate with existing Lil Nouns tools and the current bot, upgrading them to become more accurate, useful, and long-lived. Builders, developers, and vibe coders will be able to plug this data layer into their own apps and experiments.

The budget is 5 ETH for a five-month build (Sep 2025 - Jan 2026) followed by operations, maintenance, and upgrades through the end of 2026. The roadmap is structured, but steps and focus areas may be adjusted based on DAO priorities at the time, ensuring the project aligns with the community’s evolving needs.

---

Problem

The Lil Nouns AI Agent shows promise, but it hallucinates. The issue is not the model, it is missing context. DAO knowledge is scattered across proposals, Discord threads, shows, and repos. Without structured data, AI will always guess.

The deeper problem is that Lil Nouns data is not AI-native. For communities and DAOs, being legible to AI is no longer optional. Governance, culture, and dev resources must be structured, queryable, and portable, secured with the right mix of off-chain efficiency and on-chain verifiability.

---

Solution

This project builds AI-native data infrastructure for Lil Nouns and integrates it with tools already in use.

• Governance data: proposals, votes, treasury flows, auctions, secured with on-chain storage for provenance.
• Community data: Discord, Farcaster, Mirror/Paragraph pipelines with anonymization, primarily stored off-chain for scale.
• Cultural and media data: ingestion of video and audio content such as the Pirate Show and selectively the Lil Nouns Show. These are examples only. The infra will be built to handle any similar media source including podcasts, Twitter/X Spaces, community calls, recorded workshops, and future shows. Stored off-chain for practicality, with important metadata anchored on-chain.
• Developer data: ABIs, repos, upgrade histories structured for builders, with schema references stored on-chain.
• MCP server: makes Lil Nouns data portable across AI systems.
• Existing bots and tools integration: upgrading the current AI Agent and DAO-facing bots to use the same infra, eliminating hallucinations and extending their capabilities.

Technical Approach

• Ingestion pipelines normalize raw inputs into structured and semantic formats.
• Vector databases with retrieval-augmented generation (RAG) ground answers.
• Hybrid storage: off-chain for media, embeddings, and scale; on-chain for governance, schema, and metadata.
• Tiered access controls give holders advanced features while keeping core public access.
• Monitoring and feedback loops detect hallucinations and improve reliability.
• Integration hooks ensure existing bots and future tools all share the same data backbone.
• Maintenance ensures pipelines, storage, and AI retrieval remain reliable across the lifecycle, not just during initial build.

---

Budget

The ask is 5 ETH for a five-month build and operations with maintenance through end of 2026. This includes development effort, AI usage, infrastructure, monitoring, and maintenance.

Breakdown

• Infra and hosting (1.8 ETH): off-chain databases (structured and vector), storage for transcripts and media, compute for ingestion and inference, plus on-chain costs for governance and metadata anchoring. Covers scaling and monitoring through 2026.
• Governance pipelines (0.6 ETH): development, maintenance, indexing, and on-chain storage of proposals, votes, treasury, and auctions.
• Discord integration (0.6 ETH): development, maintenance, anonymization, semantic indexing, stored off-chain with metadata anchors.
• Social media ingestion (0.6 ETH): development, maintenance, connectors for Farcaster, Mirror/Paragraph, limited Twitter.
• Media ingestion (0.4 ETH): development, maintenance, and handling of audio and video sources such as Pirate Show, Lil Nouns Show, podcasts, Twitter/X Spaces, community calls, and future DAO media. Stored off-chain with key references on-chain.
• Developer data structuring (0.4 ETH): development and maintenance of ABIs, repos, upgrades structured for dev queries, schemas anchored on-chain.
• Farcaster Mini App (0.3 ETH): development, maintenance, and lightweight interface with tiered access.
• MCP server (0.6 ETH): development, maintenance, and infra for portable access.
• Integration with existing bots and tools (0.6 ETH): development, maintenance, and upgrades of the current AI Agent and DAO bots to use shared infra.
• Flex and innovations (0.1 ETH): buffer for unforeseen costs or pipeline extensions.

Budget Philosophy

I always design budgets to be tight but realistic. The amounts listed are carefully scoped to cover actual development and infra needs without padding. My track record is that I deliver what I promise, and this project will be no exception.

Open-source note: All software from this project will be released as open-source under a permissive license (Apache 2.0). This ensures transparency, long-term accessibility, and the freedom for other builders to extend or repurpose the work without lock-in.

---

Timeline

• Month 1 (Sep 2025): Infra expansion, Discord and social ingestion, begin integration with existing Agent.
• Month 2 (Oct 2025): Governance pipelines with on-chain anchoring, developer data structuring.
• Month 3 (Nov 2025): Media ingestion pipelines (Pirate Show, Lil Nouns Show, podcasts, Spaces, community calls), tiered access implementation.
• Month 4 (Dec 2025): MCP server development and testing, continue DAO bot and tool integration.
• Month 5 (Jan 2026): Farcaster Mini App launch, full MCP server deployment, system stabilization.
• Months 6–16 (Feb – Dec 2026): Operations, monitoring, maintenance, integration upgrades, hallucination reduction, incremental improvements.

Flexibility note: While this is the planned sequence, development steps may be reprioritized in collaboration with the DAO to reflect community needs at the time.

---

Outcome

By early 2026:

• The current AI Agent and DAO bots are upgraded to use grounded data, with no more hallucinations.
• Governance, community, cultural, and developer data are structured and queryable.
• Media content including shows, podcasts, Spaces, and calls are preserved and searchable.
• Lil Nouns gains a living, portable memory of its history and culture.
• Hybrid storage ensures scalability and permanence: bulk data off-chain for efficiency, critical metadata and governance on-chain for trust and provenance.
• MCP server ensures Lil data is not locked in one app but queryable by any AI system.
• Developers and vibe coders can build apps, bots, and cultural projects directly on Lil Nouns.
• All software released openly under Apache 2.0 ensures the infra remains usable, forkable, and auditable by anyone.

The real outcome: Lil Nouns becomes AI-native. Its knowledge is not fragmented, but structured, queryable, portable, and actively maintained. The infra does not just provide tools for the present, it preserves DAO culture as an AI-native archive, ensuring Lil Nouns history and creativity are accessible for years to come.

For 5 ETH, the DAO secures five months of build and nearly a year of operations with active maintenance, upgrading Lil Nouns from an experimental demo to DAO-grade AI infrastructure: reliable, extensible, open-source, and future-proof through 2026, with flexibility to follow DAO priorities.