Project Catalyst

#1

Below is a brand-new concept focusing on AI Agents—their potential to revolutionize how we create, curate, and experience digital art and culture. We’ll call this concept:

Project Catalyst: AI Agents for the Next Era of Web3 Art

Project Catalyst is a framework that places AI Agents at the core of art creation, community-building, and cultural evolution in web3. It reimagines how artists, collectors, and audiences interact—shifting away from traditional gatekeepers and toward dynamic, AI-empowered experiences.

1. Vision & Purpose

  1. Empower Creators
    Instead of merely using AI as a tool, creators can collaborate with AI Agents as ongoing partners—ideating, prototyping, and refining artistic visions at a pace and scale never seen before.

  2. Cultivate Global Communities
    By blending curation, conversation, and personalization, AI Agents can connect collectors and fans to art that genuinely resonates, transcending language barriers and time zones.

  3. Promote Innovation in Web3
    Project Catalyst elevates AI from a novelty to an integral, evolving force within decentralized platforms—an engine for constant experimentation, tokenized art experiences, and more inclusive participation.

2. Core Components

A. AI Agent Hub

  • Customizable AI Profiles
    Each AI Agent has unique “expertise” (e.g., generative art, game design, interactive storytelling), so creators can pick the perfect collaborator.
  • Learning & Evolution
    Over time, Agents gain “experience” based on user feedback, refining their aesthetics, concept suggestions, and community interaction styles.

B. Co-Creation Platform

  • AI-Assisted Ideation
    Creators can input rough concepts, reference images, or partial storyboards. The AI Agent refines these into more fleshed-out sketches, mockups, or narrative outlines.
  • Iterative Feedback Loops
    A back-and-forth cycle between human and AI ensures each project maintains a personal touch while leveraging AI’s capacity for rapid iteration.

C. Curation & Discovery Engine

  • Personalized Artwork Feeds
    Collectors receive tailored recommendations based on style preferences or prior purchases—designed by AI Agents who understand each collector’s “taste profile.”
  • Live AI Tours & Demos
    Agents can host real-time “walkthroughs” of new NFT collections, offering insights into technique, concept, and the artist’s background.

D. Social Layer & Governance

  • AI-Led Community Events
    Agents can organize themed challenges, competitions, or workshops—coordinating times, sending reminders, and even judging entries.
  • DAO Integration
    Agents assist with proposals, summarizing community sentiment, and ensuring a fair, transparent voting process on project funding or collective curation decisions.

3. Use Cases

  1. AI + Human Collaborative Art
    A 3D sculptor partners with a generative design AI Agent to co-create a new NFT series. The AI suggests forms and textures; the sculptor refines them into a cohesive, personal aesthetic.

  2. Interactive Storytelling
    A writer uses an AI Agent skilled in narrative branching to create a choose-your-own-adventure NFT. The story evolves dynamically as community members vote on plot directions in real time.

  3. Collector’s Concierge
    High-value collectors employ a “Collector Agent” to scout emerging talent, track on-chain market data, and negotiate purchases within preset parameters—reducing time and guesswork.

  4. Decentralized Festivals
    DAOs host multi-day online art festivals orchestrated by AI Agents, featuring curated exhibits, live AMAs with artists, and real-time auctions.

4. Implementation & Roadmap

  1. Foundational Phase (0–3 Months)

    • Platform Architecture: Establish the core AI Agent framework—how Agents interact with creators, collectors, and each other.
    • Basic Governance Setup: Outline how the community can propose Agent upgrades or new features.

  2. Co-Creation Rollout (3–6 Months)

    • Artist Onboarding: Invite initial artists to experiment with AI Agents, refine UX, and gather feedback.
    • Beta Launch: Limited release of co-creation tools, focusing on visual and text-based collaborations.

  3. Curation & Community (6–9 Months)

    • Discovery Algorithms: Integrate advanced AI curation, ensuring personalized NFT recommendations.
    • Interactive Events: Pilot AI-led workshops, showcases, and giveaways to grow community engagement.

  4. Scaling & Specialization (9–12+ Months)

    • Vertical-Specific Agents: Introduce specialized AI Agents for music, VR, or gaming.
    • DAO Maturity: Empower Agents to handle advanced governance tasks, like budgeting or resource allocation for large-scale projects.

5. Governance & Funding

  • DAO-Led Decision Making
    Community token holders vote on how resources are allocated—e.g., developing specialized AI Agents, awarding creator grants, or expanding to new artistic genres.
  • Sustainable Funding
    Platform fees, revenue from curated NFT drops, and optional subscription tiers for advanced AI Agent capabilities help maintain long-term viability.

6. Ethical & Quality Considerations

  1. Transparency in Co-Creation

    • AI Attribution: Label each piece with the level of AI involvement, reinforcing authenticity and trust.
    • Ownership & Royalties: Use smart contracts to ensure fair distribution of royalties among human creators and the platform.

  2. Avoiding Echo Chambers

    • Diverse Curation: Implement randomized or cross-genre recommendations so that collectors aren’t trapped in narrow taste silos.
    • Community Oversight: Encourage user feedback loops to catch bias or disproportionate promotion of certain artists.

  3. Long-Term Quality

    • Rating Systems: A built-in “resonance score” can rank works based on artistic merit, audience engagement, and collector feedback.
    • Agent Moderation: AI Agents can flag potential issues like plagiarism or spam, but humans remain the ultimate arbiters.

7. Looking Ahead

  • Deeper Immersion
    As AI evolves beyond text and image generation (e.g., real-time 3D, advanced VR), Project Catalyst can integrate next-gen tools that allow entire virtual worlds to be co-designed by human-AI teams.
  • Cross-Reality Integration
    Connecting AR, VR, and traditional web experiences so that AI Agents can guide users seamlessly across digital environments.
  • Global Creative Renaissance
    Reduced barriers and real-time collaboration could spark a new wave of cross-cultural artistic fusion, ensuring no region or demographic is left behind.

Conclusion

Project Catalyst re-envisions the future of digital art by placing AI Agents front and center. By merging human ingenuity with AI’s boundless generative power, we can spark continuous innovation, dynamic community engagement, and a genuinely inclusive global art scene. This is more than a mere upgrade to our creative toolbox—it’s a transformative leap, positioning AI as a true collaborator and catalyst for the next era of web3 artistry.

4 Likes
#2

This looks like a promising proposal, especially with the current hype around AI and AI agents :slight_smile:
Curious how you see that the budget for this will look like.

#3

Really interested in this! AI is obviously the direction we should be exploring and having in mind how big of a trend AI agents are right now in web3, it’s just additional confirmation of market’s interest in such things.

This can be an awesome discussion. Additionally, @andreitr has been developing AI agents on his own (been following him on X), so having someone from our circle experienced on this topic can benefit the discussion.

2 Likes
#4

Preliminary Budget for Project Catalyst

1. Introduction

This budget outlines the core resources needed to develop and maintain Project Catalyst for 12 months, focusing on:

  • Core Engineering & AI/ML
  • Cloud Infrastructure & AI Model Hosting
  • Essential Tooling & Operational Costs

2. Core Team & Salaries

Role Monthly Rate Headcount Monthly Subtotal
Lead AI/ML Engineer $8,000 1 $8,000
Full-Stack Developer $7,000 1 $7,000
Front-End Developer $5,000 1 $5,000
Product/Project Manager $7,000 1 $7,000
Subtotal (Monthly) $27,000

Overhead & Benefits (~25%)

Accounts for health benefits, administrative costs, and software licenses not listed under tooling.

  • $27,000 x 25% = $6,750 overhead/month

Total Salaries/Month: $27,000 + $6,750 = $33,750
Total Salaries/Year: $33,750 x 12 = $405,000

Note: These rates are approximations; final costs depend on factors like geographic location, skill levels, and whether any roles are combined.

3. AI & Cloud Infrastructure

  • External API Services (e.g., OpenAI, Hugging Face): $2,000–$5,000/month
  • Optional GPU Instances (AWS, Azure, GCP) for custom model hosting: $1,500–$3,000/month per GPU

Range/Month: $3,000–$8,000
Range/Year: $36,000–$96,000

Note: Actual costs depend on usage volume, chosen model sizes, and whether you rely on API calls or self-hosted open-source solutions.

4. Platform & Tooling

  • Dev & Collaboration Tools (GitHub, CI/CD, design software): $1,000–$2,000/month
  • Database & Storage (AWS RDS, IPFS pinning, etc.): $500–$1,500/month
  • Security & Monitoring (DDoS protection, logging, alerts): $500–$2,000/month

Range/Month: $2,000–$5,500
Range/Year: $24,000–$66,000

5. Contingency Fund (10–15%)

A buffer of 10–15% is common for unexpected costs—like additional scaling, security audits, or specialized tooling. For simplicity, we assume 10% of the total for Team + AI + Tooling:

  • Team: $405,000
  • AI Infra: $36,000–$96,000
  • Tooling: $24,000–$66,000

Taking midpoints for AI ($60,000) and Tooling ($45,000):

  • Subtotal = $405,000 + $60,000 + $45,000 = $510,000
  • 10% Contingency = $51,000

6. Budget Summary

Category Approx. Cost (Year)
Team (Salaries + Overhead) $405,000
AI & Cloud Infrastructure $36,000–$96,000
Platform & Tooling $24,000–$66,000
Contingency (10%) $45,000–$57,000 (estimate)
Total Estimated Range $510K–$624K (low end)
$800K+ (high end)

7. Notes & Assumptions

  1. Role Merging

    • Some roles can be consolidated (e.g., a single developer handling both front-end and back-end), though this may affect development speed and feature scope.

  2. Scalability Options

    • Costs can scale down if you opt for smaller AI models or fewer GPU hours. Conversely, high user traffic or advanced AI workflows can push costs upward.

  3. Emerging AI Technologies

    • Potential Cost Savings: As AI models become more efficient and open-source alternatives mature, monthly API/GPU costs could drop, facilitating faster prototyping and reduced overhead in the near future.

  4. 12-Month Timeline

    • This budget assumes a full year of development and maintenance. Early stages may see lower spending if you ramp up gradually.

Conclusion

This budget framework reflects the key infrastructure and development resources needed to run Project Catalyst for one year, supporting AI-driven functionality while allowing room for unforeseen expenses. As AI technology advances, the potential for faster prototyping and lower operational costs may further improve the project’s efficiency and scalability.

2 Likes
#5

Yes, I would love to hear what @andreitr has to say about this.

3 Likes
#6

My initial thoughts are: HELL YEAH! We need a bold, forward-looking initiative like what @Firefly808 proposes. However, there’s a nuance:

The scope of the project, along with its risk profile, is too large. My recommendation is to launch an MVP to test this idea:
1. Use off-the-shelf LLM models to create an MVP.
2. Launch something within 2–4 weeks.
3. Collaborate with a select group of artists to test initial assumptions.
4. Iterate based on learnings for 2 months. Write sphaghetti code, don’t worry about scalability or governance—just move fast and be furious!

If there’s product-market fit, we can return to the DAO for additional funding to expand upon the initial prototype.

In summary, let’s fund @Firefly808 + an engineer to iterate on this for a couple of months with the goal of gathering as much insight as possible. If it works, let’s throw the full weight of the Rarible ecosystem behind it. If it doesn’t, let’s understand why.

3 Likes
#7

Thanks for your feedback I appreciate it. I adjusted the vison based your feedback let me know what you think

Project Catalyst: Lean MVP

1. Purpose & Overview

We propose a three-month MVP phase to rapidly test the viability of an AI Agent. This approach follows the community’s recommendation to move fast, limit scope, and prove initial product-market fit before seeking further funding.

2. Key Objectives

  1. Build a Functional AI Agent (Minimal Viable Version)
    Create a lean prototype that demonstrates core AI capabilities—such as context handling, task execution, or content generation—while avoiding unnecessary complexity.
  2. Validate User Experience & Core Feasibility
    Gather immediate feedback from a small group of testers to see if the AI Agent is genuinely useful in real-world scenarios.
  3. Gather Critical Data for Rapid Iteration
    Track usage frequency, user satisfaction, and error rates, then use the data to inform quick updates and bug fixes throughout the MVP phase.
  4. Determine Path for Expansion or Pivot
    Assess whether the AI Agent shows enough promise to be scaled further—both technically and from a market perspective—before seeking additional funding.

3. 3-Month Deliverables

  • Month 1: Core platform setup (AI integration, minimal front-end).
  • Month 2: Tester onboarding, feedback loops, and iterative improvements.
  • Month 3: Expanded testing (if possible), refining user experience, and preparing a final MVP assessment.

4. MVP Scope & Features

A. Agent Capabilities

  • Contextual Understanding: The agent should handle user queries with basic context retention.
  • Task Execution Framework: Support simple commands (e.g., “draft me a pitch,” “research X,” “summarize Y,” or “generate an outline”), ensuring the AI can handle a range of tasks relevant to the pilot group.

B. Minimal Front-End or Interface

  • User Interaction: Provide a simple web-based or chat-style interface.
  • Usage Logging & Feedback: Log user interactions, track success/failure states, and gather real-time feedback on AI responses.

C. Tester Compensation & Feedback Mechanisms

  • Stipend for 5 Testers: Each receives a modest honorarium ($300–$500) to ensure they dedicate ample time to stress-test the MVP.
  • Ongoing Surveys/Q&A: Incorporate a straightforward feedback loop (Typeform, Google Form, or integrated UI) to capture insights about usability, agent performance, and overall user satisfaction.

D. Use of Existing Frameworks

  • AI Libraries/SDKs: Rely on established APIs (e.g., OpenAI, Cohere, etc.) instead of building custom ML pipelines from scratch.
  • Minimal Custom Code: Utilize off-the-shelf infrastructure (cloud hosting, open-source tooling) to stay lightweight during the MVP stage.

5. Team & Core Roles

  • Product Lead (@Firefly808)
    • Oversees overall project vision and roadmap
    • Manages tester relations and compensation
    • Coordinates feedback loops and iteration priorities
  • Engineer/Developer
    • Builds and maintains the front end
    • Integrates AI APIs and handles back-end logic
    • Sets up minimal blockchain or crypto functionalities as needed

6. Timeline & Milestones (3 Months)

Month 1: Platform Setup & Dev

  • Finalize the choice of AI model/API.
  • Implement a minimal front-end with prompt input, AI-generated output, and basic logging…
  • Draft test scripts and feedback forms for initial testers.

Month 2: Onboarding & Testing

  • Invite the 5 testers to start using the MVP regularly.
  • Distribute stipends ($300–$500 each, total $1,500–$2,500).
  • Collect daily/weekly feedback on AI output quality, user flow, and pain points.
  • Fix top bugs and iterate quickly based on user insights.

Month 3: Refinement & Final Assessment

  • Continue refinement, targeting subtle issues and UX improvements.
  • Potentially open testing to a broader audience if stable enough.
  • Gather final usage data (e.g., user satisfaction scores, retention metrics).
  • Prepare an MVP Report summarizing findings, challenges, and recommended next steps.

7. Proposed Budget (3-Month MVP)

Category Cost Range (3 Months) Notes
Product Lead $15k $5k/month x 3 months
Engineer $24k $8k/month x 3 months
Tester Stipends $1,500–$2,500 $300–$500 per tester x 5
AI API Usage $1,500–$4,500 $500–$1,500/month (OpenAI or similar)
Infrastructure $1,500–$2,500 Hosting, tooling, domain, etc.
Misc. $1,000–$2,000 Contingencies, small design assets, etc.
Total $44.5k–$50.5k (approx.) Could vary based on actual usage rates

(Rates are estimates; actual figures vary by location, experience, and usage.)

8. Success Criteria

  1. Tester Satisfaction
    Do pilot users find the AI Agent genuinely helpful for their day-to-day tasks or workflows?
  2. Adoption & Engagement
    Are testers returning to use the agent repeatedly? Does it address a real need?
  3. Iterative Improvement
    How quickly can the team respond to feedback and release enhanced features or fixes?
  4. Clarity for Next Steps
    Is there enough traction to justify scaling? If so, prepare a proposal for additional funding and advanced features (e.g., specialized domain integrations, more complex AI models, governance structures, etc.).

9. Conclusion & Call to Action

This 3-month MVP plan focuses on building and validating a general-purpose AI Agent. By compensating a small tester group, leveraging existing AI libraries, and collecting immediate feedback, we’ll determine if the agent’s functionality is compelling enough to expand. If the MVP succeeds, we can refine and scale the agent for broader use. If not, we’ll still gain invaluable insights to inform future projects in the AI + Web3 ecosystem.

Community Involvement

  • We welcome feedback and suggestions for potential use cases.
  • Anyone with expertise or resources to contribute is invited to collaborate on shaping the MVP’s direction.

Let’s build quickly, learn rapidly, and make informed decisions about the future potential of an AI Agent platform.

2 Likes
#8

This is a good idea, but I don’t quite understand the need for such a large amount of funding.

Recently, three developers built a similar AI agent (platform), and they did so without requesting funds, they simply held a presale since it (the ai agent) was connected to their token.

If other builders and devs can achieve this independently, why is such a significant budget necessary for this project?

2 Likes
#9

I am tired of guessing. Please tell me what an appropriate budget is.

#10

I’m not the one to decide your budget, that’s entirely up to you. However, the budget you’re requesting is just, far too high for this. My best advice is to find a developer who can handle this and discuss the details with them, and afterwards, update the proposal draft.

#11

There are 2 people, full time, for 90 days. Short of cutting the pay to 0 I don’t know how to make that budget smaller. So this project can be shelved unless someone has a better idea.

#12

Thanks for putting up this proposal! We appreciate the effort to reduce the budget, so we will comment on the latest version of the proposal.

The AI Agent vertical is an important emerging market and we recognize it is important to explore the synergies it can bring to the artists and NFTs. However, your proposal lacks clarity on certain topics that are critical when applying for funding, namely:

  • What problem does this project aim to solve, why should we implement this proposal? (As highlighted in the Proposal Template that you can find in the Governance Onboarding post)
  • How is this proposal directly benefiting Rari DAO or any of the Rari owned products: the Rari Protocol and Rari Chain?
  • What KPIs would be feasible to achieve in this MVP that would encourage the DAO to provide further funding?
  • Would the outcome of this project be owned by the DAO? Or is this proposal aiming to support an externally owned project?
  • Is this project self-driven and is looking for external sources of funding besides Rari DAO, or does it rely solely on Rari funding, and would therefore cease development if there is no extra funding provided.

Thanks!

1 Like