Abstract
MoonDAO is currently behind schedule on its 2030 goals. To accelerate development and improve operational efficiency, this proposal advocates for the adoption of next-gen AI frameworks for autonomous agents. These agents will streamline our processes, reduce human error, and enhance transparency, ultimately accelerating development towards our objectives. Before we take our first baby steps into this domain, we must first prioritize building a solid floor to walk on. This proposal aims to build that floor by focusing on a long-term, platform-agnostic vision of our AI strategy.
Problem
Our operational data, while publicly available, is complex and challenging to interpret. This complexity leads to processes that are error-prone, slow, and inefficient. For instance, the quarterly incentives process is so intricate that it requires three executives working independently to validate the results. This process of calculating and coordinating reward distribution can take days, often dragging on for weeks or even months.
These kinds of inefficiencies not only hinder progress but also divert valuable resources from more strategic tasks. Additionally, the reliance on executive oversight centralizes operational control, which runs against our goal of progressive decentralization. The rapid pace of technological advancement further complicates our situation, creating uncertainty around the longevity and relevance of any given tool, thus making long-term planning a challenge, but a challenge worth tackling.
Solution
To address these challenges, we propose leveraging new framework technologies that enable digital agents to make informed decisions, reason through complex queries, and track important details. With recent developments, these agents are now capable of handling routine tasks that are easy to explain in natural language or code, and importantly, can be undone if needed. This means they operate proactively and without human intervention. This approach will save time, reduce human error, and significantly increase operational transparency, freeing up valuable hours for creative and strategic work.
Our solution is tech-agnostic, grounded in graph database modeling, which will allow us to focus on our strategy first and choose the appropriate framework technology to match our plan. Unlike traditional databases, graph databases excel in handling complex contextual relationships between entities, making them ideal for two key aspects:
- Knowledge Architecture: This represents the agent’s working memory, storing data in a way that both LLMs and humans can easily understand and reason about. This will enable us to maintain providence of information and essentially make the agent hallucination-proof.
- Cognitive Architecture: This involves mapping operational workflows and dependencies, encompassing all interfaces the agent will engage with (Discord channels, DMs, Coordinape, etc.). With the benefit of the knowledge architecture, the agent will have complete contextual understanding on how we use these platforms.
By basing both the knowledge and cognitive frameworks in a graph format, we can build a foundational strategy that can adapt to emerging technologies and developments. An additional long-term benefit of this approach is that it allows for the agent’s inner workings to be scrutinized by even non-technical members, and opens the opportunity to put it under control of community governance, promoting transparency and trust.
While this project has ambitious goals for streamlining operations across MoonDAO, our initial focus will be on automating the community circle to demonstrate immediate value for the community.
Benefits
Time Cost Savings This benefit will be the most immediate and impactful. Currently, the operation of the community circle is often overlooked as other operations, such as incentive coordination, take priority. Automation will ensure that these tasks are handled efficiently, allowing us to grow the community circle much faster. We anticipate a 50% reduction in time spent on community circle operations.
Reduced Human Error Potential Initially, the system may require close monitoring. However, as the agent’s decision-making process will be transparent and traceable, the system should become more robust and trustworthy over time, significantly reducing the potential for human oversight errors.
Decentralization of Executive Control Graph databases, with their inherent contextual understanding, are the ideal information store for agents. This step towards decentralization, while outside the immediate scope of this project, aligns with MoonDAO’s broader objectives.
Risks
Rapid Tech Obsolescence
The risk of technology becoming obsolete is almost certain to manifest. Our graph database modeling approach abstracts out specific technologies, allowing us to adapt as needed while keeping the focus on our strategic objectives. In effect, the implementation of technology can be built directly from a platform-agnostic map of our knowledge and workflows.
Reduced Oversight
As responsibilities are handed off to digital agents, ensuring these systems are trustworthy and auditable will be vital. Graph databases like Neo4J offer query languages that can support this, but it will be crucial to ensure that the agent does not make any final or irreversible decisions without human review.
Objectives
Objective #1: Automation of the initial phases of our content development strategy Key Results for Objective #1:
- Automated aggregation and summarization of space news
- Automated publication in CSV format
- Automated notification through discord
Objective #2: Automation of community circle coordination Key Results for Objective #2:
- Automated handling of circle configuration
- Automated engagement, notifications, and info collection
- Automated creation and publication of reports and visualizations
Member(s) responsible for OKR and their role:
- Mitchie, Lead Data Architect
- Colin, Lead Developer
- Ryan, Consultation
- Pablo, Consultation
Deadline: End of Q4, 2024
--- nance-actions
- type: Request Budget
payload:
projectTeam:
- discordUsername: Mitchie
isRocketeer: true
discordUserId: "926244811177066558"
payoutAddress: "0xb1d4c1B9c8DA3191Fdb515Fa7AdeC3D41D014F4f"
votingAddress: "0xb1d4c1B9c8DA3191Fdb515Fa7AdeC3D41D014F4f"
multisigTeam:
- discordUsername: pmoncada
discordUserId: "494739213137346561"
address: "0x679d87D8640e66778c3419D164998E720D7495f6"
- discordUsername: name.get
discordUserId: "760130086048366663"
address: "0x80581C6e88Ce00095F85cdf24bB760f16d6eC0D6"
- discordUsername: Mitchie
discordUserId: "926244811177066558"
address: "0xb1d4c1B9c8DA3191Fdb515Fa7AdeC3D41D014F4f"
- discordUsername: ryand2d
discordUserId: "842541430186573894"
address: "0x78176eaabcb3255e898079dc67428e15149cdc99"
- discordUsername: ""
discordUserId: ""
address: null
budget: []