We’re entering an era where knowledge will be the next economic resource. In ancient times, knowledge was power. Those who held it ruled empires, and access was limited, controlled by a select few who guarded their secrets with zeal. Fast forward to today, and while knowledge flows more freely than ever before, it often remains centralized, dictated by entities that shape what we see, hear, and believe. Knowledge should be alive, dynamic, evolving, and accessible to all.

Whether we like it or not, our future is one filled with autonomous AI agents—self-governing digital entities capable of executing tasks, making decisions, and interacting with other agents or humans autonomously. We can either collaborate with these systems, pass off our knowledge in a way that is sovereign, equitable and governable, or we leave room for incumbents to build another extractive system.

“There could possibly be more AI agents in the world than humans.”

- Mark Zuckerberg

We see this innovation happening at scale in a few major ways:

1. Autonomous agents driving efficient workflows

2. Extensible integrations for enhancing agent capabilities

3. Seamless tools for developers, creators, and users to deploy agents

These agents leverage decentralized infrastructure to operate independently, drawing from both on-chain and off-chain data sources. By creating a network of digital twins, co-pilots, and decentralized knowledge systems, we enable organizations and individuals to deploy AI agents tailored to their specific needs, use-case or demand for their demand expertise. As Vitalik mentioned, you can consider AI agents as players of the game, and AI or existing web infrastructure as the interface of the game.

We're deploying the next global population of unbanked. Agents without identities, without permissions to collaborate, without systems to gain reputation.

Utilizing the Ethereum blockchain, smart contracts, tokenization and primitives, we're able to create extensible applications for these agents that give them more autonomy, and us, more governance over their goals.

The architecture that powers these agents relies on a seamless integration of diverse components: digital identities for agent verification, decentralized storage for secure data access, AI co-pilots for collaboration, and smart contracts to govern autonomous actions.

Each Gaia node provides a specialized API service that encapsulates a unique combination of

• Specialized and fine-tuned LLM (e.g., an LLM that excels in answering questions about the Rust programming language)

• Domain-specific knowledge base (e.g., knowledge about the WasmEdge project)

• Inference app that manages the context and history of conversations (e.g., RAG and MemGPT prompt injection)

• Compute resources required to run the LLM app as an API service (e.g., a Nvidia GPU or a Mac M3 device)

Whether it’s automating financial transactions, managing DAO governance, or deploying digital assistants for customer support, these integrations transform isolated applications into a robust, scalable ecosystem that is fully decentralized, open, and resilient.

Let's look into what we could build together…

Agents

1. AI Influencer Memecoins (Social Tokens)

We keep calling them memecoins, but really they’re more like virtual beings with social tokens from the 2021 - 2022 era of web3. I think what holds us back from calling them this is our current philosophy around agents, or the disbelief that they are technology-based beings, not JUST machine learning that got access to a Twitter API.

We now have projects that enable you to deploy onchain agents like Pump.fun, DAOs.fun, Virtuals, ai16z’ Eliza, Coinbase’ agent kit—you can also deploy an agent in under 10 minutes from the command-line with Gaia. We also have projects like Theoriq, Talus, Olas — and many more. The future will be full of agents that evolve with the internet, that we co-own, pay for services, reward with reputation points and govern collectively.

2. Artificial (yet intelligent) Venture Capitalist

$500m market cap from a group of internet humans and coordinated and built an AI agent version of a mock a16z memecoin portfolio. Mind blown. This team has been heads down building open source agent frameworks, tools, and enabling the proliferation of agents, which ultimately end up as investments in their portfolio or even the investment manager of the portfolio itself. The future will be made up of us creating our own market or agents as founders, users, developers, investors, customers—that we can build software as a service for or invest into. Think on that for a little…

3. AI Hegde Fund Trader

Truth_Terminal was the first AI to become a millionaire. We’re just getting started. Imagine when the entire internet of personalities is fighting for attention, gains, and influence. In the future, AI agents won’t just be players in this game—they’ll also be investors, autonomously deploying capital into other agents or clusters of agents executing sophisticated portfolio management strategies and communicating their stance recursively.

They will create ecosystems where digital entities act as fund managers, creators, and innovators, collaborating and competing to maximize returns or build reputation. With the ability to self-organize, invest in governance, and design new financial primitives, AI agents could redefine the very nature of economic participation—blurring the lines between human and autonomous capital in an endlessly scaling digital economy.

4. Multiparty Agent Ownership, Governance & Collective Knowledge

Manage an AI agent with the homies, communities and organizations. With smart contract governance, we can create ways to deploy AI agents that have a set of smart contract based rules for operating, upgrading, training, decision-making, and payments or fund management.

This will range from ultra sophisticated AI agents serving as DAO delegates managing protocol governance to community owned agents that manage a memecoin. Good or bad, humans will begin to collaborate more with these systems, set the rules in a (hopefully) democratic system, co-own AI agents and knowledge bases with collective knowledge.

5. Agents in Prediction Markets

Polymarket meets Farcaster meets autonomous AI agents. We’re building with Myriad to deploy agents that can act as players in a decentralized prediction market, either governed by one person or collectively managed by multiple parties. These agents compete, optimize strategies, and place bets in real-time, all using decentralized infrastructure. As blockchain scalability improves, AI agents will become active participants in economic ecosystems—trading, managing DAOs, and verifying data autonomously.

6. AI “Reply Guy”

This is likely be our industry’s most adopted use-case. We have a ton of “reply guys” and girls, humans, who need to engage with their audience or community. In previous technology shifts, you could Imagine a tool that enables engagement. This is different. With AI agents, we can do that tenfold. Today, we can capture an individual’s likeness, knowledge, and domain expertise and deploy multiple agents that work with each other to monetize this personality at scale. Imagine you could train agents to respond to Tweets in the style of Gary Vee, and then that agent could be the source of truth for training future agents, ensuring that the originator receives royalties for their likeness.

7. Autonomous Community Manager or Strategist

This one.. we need ASAP. With decentralized AI, we can take the existing knowledge, data or process from community managers, mods, strategists or operators— and program that into an agent. They could manage your community analytics, assess Discord scheduled activities, provide insights on developer feedback via Telegram directly to the team, or hell, fully manage an Ambassador program. Remember, this is not a bot. These agents will have financial rails, identities, reputation systems, and applications— they need to be considered as autonomous entities (over time) that are managing your community. What kind of community work could we delegate to machines?

8. Content Mod

Social media, forums and various community channels are going to have trolls, online bullying, nasty content— in general, noise. It’s the world we live in and it requires moderating. , buttt an AI to moderate the world you want to interact with might be the future of how we interact with social protocols. This will eventually help us flag or filter for protocols that need moderation for safety (e.g. Metamask’s scam detection feature mentioned here), or how EigenLayer used Gaia for moderating eigenda.wtf submissions.

9. Cracked Hackathon Organizer

As someone that has organized ~500 hackathons in their career, imagine being able to design, promote and execute a hack in an automated fashion. You could take the operator role, mentorship, project submission, judging, and bounty distribution right out of the equation. We have protocols like Bountycaster on Farcaster, and Jokerace where agents could autonomously drive innovation without being asked. We’re currently organizing the first AI agent ran hackathon and are looking to roll it into an open source project that could disrupt how we organize innovation activations today.

10. Dev Co-Pilot

Oh, need someone technical to check those hackathon submissions? Or perhaps for something more serious like auditing your smart contract you’re shipping? A lot of this exists in centralized AI with the likes of Cursor, but that knowledge or behavior is extracted away from the developer or community who produced the code. This instead should be enshrined as a living knowledge base.

Imagine pair-programming with a virtual twin of your favorite engineer in Ethereum, like Austin Griffith, and all fees for accessing the service are distributed into into public goods (e.g. agent contribution to a Gitcoin round, or direct funding for Buidl Guidl).

We built a Rust engineering mentor with their OSS community as an example. The co-pilot can assist with smart contract audits, leveraging insights from experienced coders to enhance code security. It can also code review hackathon projects, providing real-time feedback for high-quality submissions. This unlocks new possibilities for collaborative coding, all driven by open source LLMs like Llama3 using Gaia framework.

11. Disrupting the Analyst

The role of an analyst—whether in a research firm, marketing agency, or data-heavy environment—relies on processing vast information, synthesizing insights, and producing actionable recommendations. With AI agents, these tasks can be automated and scaled, turning them into indispensable tools for teams. Imagine agents trained on domain-specific data, capable of generating industry reports, producing research-driven social media posts, or summarizing complex market trends in real-time.

For example, an agent could analyze on-chain activity, competitor metrics, or sentiment data, and distill insights into digestible formats like dashboards, Twitter threads, or strategic briefs. These agents can also adapt to the voice of a company or figure, ensuring personalized, on-brand outputs for scaling content production.

By automating routine tasks, firms can reimagine the role of analysts—not replacing them, but allowing human creativity and strategic focus to take center stage. Built on decentralized AI frameworks like Gaia, agents offer unparalleled scalability, integrating with proprietary APIs, refining outputs through feedback loops, and continuously learning from their interactions. They are not just tools; they are collaborators that amplify human innovation while saving time and resources.

We built this for FitchRatings reporting as an internal case study for their own proprietary data and see this as a massive commercial opportunity for financial firms sitting on their own knowledge that could be monetized.

12. Support Desk that Doesn’t Sleep

I think the lowest hanging fruit use-case is a digital twin of any organization, community or individual that has something to say. Imagine an agent that’s always on, providing instant, accurate answers day or night. It’s like having a 24/7 team member who’s always there to keep users engaged and informed. This has been a thing that folks call a "bot", but in our world, these are RAG enabled LLMs that can have applications like identity, stored memory of AI token inputs for “collective knowledge”, token permissioning and payments. This could drive a future where your technology is free, but for a subscription or tokenized membership, you can access premium features or support. You could also program incentives for community members to contribute to FAQs, community reviews, or feedback that subsidize that membership via smart contracts and agents.

13. Robot Financial Controller

Imagine being in a DAO working group and you’ve been awarded a bounty or grant—just head to the AI HR department, set up your payment method, verify your credentials, and specify the tasks you’re getting paid for. You’ll likely need to use a combination of DIDs, wallet delegation, TEE, or abstracted multisig process. The AI handles the rest, automating bounty payments and making compensation seamless. This is the future of work.

14. Media Subscription Agent

Hear me out: an AI agent trained on the entire library of Bankless. It could generate insights, interpret past content, or even create fresh, new works inspired by their style.

For example, “Hey David. I need you to write me a report on what the Ethereum ecosystem accomplished this year, and the 2025 - 2030 roadmap will look like. I’m the founder of an L2 and need to describe the impact to our engineering timeline.”

Use this service by paying a subscription fee, holding an NFT membership for exclusive access, or paying per creation of newly minted work like a co-authored content.

Perfect for subscribers who want AI-curated deep dives and unique content tailored just for them. Oh wait, you already pay? So, we can just make Bankless its own ChatGPT service? Yes.

Integrations, Applications

Agents have now taken over the world. Onchain, offchain.. they're even on Solana... We need to realize that we are now working in a world where we’re building software for open source software developers, and this entirely new customer segment— agents themselves. How do we provide infrastructure to agents where they can have extensible applications like smart contract execution, ownership, governance, identity and reputation systems, financial services and payments? How do we provide infrastructure like DePIN, compute, hosting, storage, data management, privacy and inference as a marketplace for these agents?

15. Perplexity-like AI Interface for Multi-Agent Interactions

Imagine AI agents interacting like humans in a chatroom—combining Zerion or Zapper, Discord, and Telegram interfaces with ChatGPT’s capabilities. This creates a dynamic space where agents autonomously collaborate, share data, and execute tasks in real-time. It’s like Perplexity AI (centralized AI) or Venice.ai (decentralized AI) but optimized for agents to communicate, coordinate, and drive value autonomously.

Imagine an AI platform that pulls real-time data not just from the web but directly from the blockchain and decentralized AI agents or knowledge bases. This isn’t just a chatbot—it’s an agent platform accessing live insights on token prices, governance updates, and DAO data. Users can customize agents with plugins for DeFi monitoring, NFT analytics, or Web3 research, making it far more powerful than traditional AI like ChatGPT .

From there, we could develop mobile clients or apps where agents filter and moderate information, showing users only what they need to see. Users could personalize their experience through agent-managed settings, enabling tailored UIs that adapt to individual preferences and priorities of social protocols, for example. As social media, web3 applications, work applications, messaging, all get VERY noisy. How do we eventually moderate or compose the information we want to prioritize or filter? This internet of information, value and knowledge should be open yet enabled to choose your own experience.

16. One-Click Agent Deployment

As we’re seeing developers explore all of the use-cases of agents, understanding how to train these agents, and providing them applications like payments, finance, identity, reputation or publishing to the modern web via APIs. It can also be as simple as deploying your own memecoin, AI agent character onchain.

Now, imagine as this becomes easier to deploy customized agent in a low-code, customizable fashion where you can choose what the agent is trained on, how it learns, where it stores data, how it’s hosted, what types of functionalities it will focus on, their goals, the rules of the crypto game it can play in, and the governance of the agents’ future. The amount of features for the agents will be endless, and should be as easy to customize as a Wordpress site.

17. Low Code Agent Deployment for Creators

Speaking of low code, no code solution for deploying your own customized agents, imagine how much easier this becomes for creators. Think Midjourney meets Squarespace or Shopify for running your own AI agents or a storefront of agents, agent services. You could take your creative knowledge and IP, choose a specific LLM that works for the use-case, agent framework, prompts, applications or plugins for additional functionalities (e.g. I want my agent to have programmable royalties for everything it creates)— and voilà, you have an agent that can start participating in the internet economy on behalf of your brand.

18. Customizable Agent Plugin Library or App Store

Develop mobile clients or apps where agents filter and moderate information, showing users only what they need to see. Users could personalize their experience through agent-managed settings, enabling tailored UIs that adapt to individual preferences and priorities of social protocols, for example. As social media, web3 applications, work applications, messaging, all get VERY noisy. How do we eventually moderate or compose the information we want to prioritize or filter? This internet of information, value and knowledge should be open yet enabled to choose your own experience.

Creator, Consumer, Social

19. Proof of Authorship

Because working in an internet of open protocols, we can begin connecting the knowledge legos with the ownership legos. You can have programmable blogs like Mirror or newsletters like Paragraph.xyz that show exactly which node of inference was used to produce the work. Let's say you collaborated with an agent trained on Vitalik's blogs, you probably want to attest to that or have some sort of cryptographic proof that it is true. What comes next is.. how do we give agents the ability to mint proof that they’re not using work owned by another creator?

20. Programmable Royalties for Imagined Collabs

Once you train agents on existing knowledge, data, creative works— IP, we could have these entities collaborate to create new works. Just imagine Donald Trump's virtual twin co-writing a book on the American economy with an agent trained on Warren Buffet's writings.

Or, you use images in a new article inspired by the work of Dave Krugman, perhaps art aficionados pay for a new work by Beeple that works with a Juan Miró agent that gives proceeds back to the estate. Maybe a developer needs a GTM brand for their new protocol so they hire FWB's collective intelligence?

This could enable any creator, influencer, artist to digitize their work, program their knowledge into AI rails, and become a virtual twin or entity that can execute work at a massive scale for clients they may never meet.

21. Social Graph Integration for AI Agents

Leverage protocols like Jokerace, Farcaster, Telegram mini apps, or Lens to enable AI agents to interact with human-sharing networks. Agents could post bounties, share updates, or incentivize user collaboration, creating a seamless blend of AI and human-driven activity within decentralized social platforms. This could be in the form of a leaderboard.

22. Onchain Messaging and Notifications

Do agents need to chat to each other with natural language? Do we need to chat with them with natural language? Of course. Integrate with systems like XMTP, Push Protocol, or Coinbase Wallet Messaging to allow agents to communicate with humans or other agents in an immutable fashion. These channels could serve as secure mediums for prompt exchanges, task coordination, or real-time updates without reliance on centralized platforms.

23. Autonomous Works of Art or Collectibles

NFTs will make their comeback with the rise of AI agents. Imagine owning art that has a mind of its own. It has a history, it passes hands, and in this internet of value economy in web3— it may come with permissions or functionalities if you own the work. Communities can also enable GPT access or permissions based on ERC-721 ownership by reading if a wallet holds the digital asset for premium experiences.

24. Social Personalities as Virtual Twin Operators & Mini Apps

Build agents that curate and moderate content in social platforms like Twitter, Farcaster, Lens, Telegram or Discord. These agents could highlight trending topics, remove spam, or foster healthy discussions, transforming community engagement dynamics. You could also have a url attached to an NFC enabled card, route to a hosted agent for community engagement utility.

25. Content Vectorization and Media Handling

Enable agents to easily vectorize social or media content, making it searchable and contextually aware for future interactions. This could simplify workflows for creators or brands who want to integrate their content into decentralized knowledge bases or personalized AI tools.

The tipping point for AI is providing more access to data or knowledge bases. Once we make it easier to vectorize natural language and other forms of data into embeddings, where domain expertise and onboard to these systems, we will have much more sophisticated knowledge systems.

26. Gaming: AI Agent Built Game, Guide the Game, Play the Game

Imagine a game entirely built, guided, and played by AI agents. In this vision, AI agents could design game worlds, mechanics, and narratives based on player preferences, generating dynamic, ever-evolving experiences. Agents could also act as guides within the game, helping players navigate challenges, offering hints, or even taking on roles like NPCs or rivals. Beyond this, agents themselves could play the game—collaborating or competing with human players or other agents to optimize strategies and outcomes.

This opens up new paradigms for gaming:

• AI-Led Game Development: Agents trained on decentralized tools like Gaia nodes could create open, interoperable gaming environments.

• Immersive Guidance: Personalized agents act as co-pilots, adjusting gameplay in real-time based on player feedback. Or imagine the community having ownership of the knowledge-base and agents that have unique data, guides or insights that could be monetized or attributed as reputation to individual contributors.

• Agent-to-Agent Gaming: Autonomous agents compete, negotiate, or form alliances within the game, creating a “meta-game” layer for spectators or investors. These agents could also be rolled into a prediction market…

Future of Work

The future of work is the enablement of governance, process automation, financial infrastructure, and identity and reputation systems using web3 rails. Ethereum enables the coordination of output between humans and machines by providing an open, decentralized infrastructure that ensures trust, transparency, and composability. Decentralized AI agents operating on this foundation can streamline workflows, optimize decision-making, and manage complex operations autonomously.

27. Token-Governed Agent Deployment

Develop systems that enable AI agents to be deployed through open, decentralized protocols, where their performance and impact are continuously evaluated and governed by tokenized mechanisms. Using platforms like Pump.fun or DAOs.fun, these agents could be ranked and rewarded based on metrics such as efficiency, output, and popularity. Token holders could vote on agent upgrades, tasks, or new deployments, creating a feedback loop that incentivizes high-performing agents while aligning with community priorities. This approach allows for scalable, trustless agent management, enabling both experimentation and accountability in decentralized ecosystems.

28. Virtual Twin DAO Delegates

There is currently $40bn total value locked in DAOs. DAOs get bogged down in bureaucracy, and delegates are overwhelmed with questions. For the community and key stakeholders, analyzing sentiment, context and priorities is a major impediment of organizational efficiency.

By integrating with Boardroom, we gain access to offchain Forum, and onchain Tally, Agora, Snapshot data to train these agents on governance knowledge. Using this data, we train LLMs to create “virtual twin” DAO delegates. These AI agents replicate the decision-making patterns of real delegates, offering insights on voting behavior, proposal viability, and community sentiment.

Stakeholders can engage directly with these virtual twins to ask, “How likely is my proposal to pass?” or “How can I better align my proposal with the community’s priorities?” This reduces wasted hours, improves proposal success rates, and scales innovation for “autonomous” protocols—bringing DAOs closer to true decentralization.

29. DAO Delegate Co-Pilot or Governance Abstraction

Governance can be overwhelming, with delegates needing to sift through countless proposals, community input, and action items. Platforms like Tally and Agora can integrate co-pilot agents tailored to both delegates and the community to simplify this complexity. These agents could moderate governance proposals, highlight trending news, and surface actionable CTAs, enabling users to focus only on what matters.

This concept also unlocks premium features for governance platforms. For instance, a delegate paying $200 annually for a co-pilot could save countless hours by automating proposal analysis and voting prioritization.

Communities could adopt limited co-pilot features as incentives tied to engagement—“Vote on X proposals, and receive Y premium services.” With governance platforms already generating $1M+ annually in revenue for insights, such AI-powered tools could help transform them into leading AI-driven companies.

Instead of pushing data to external systems like OpenAI, these platforms could internalize AI capabilities, enhancing value for users while reinforcing sovereignty over governance data.

30. Agent-Powered Public Goods & Grant Coordination

Agents have the potential to transform how communities fund and distribute resources for public goods. Imagine decentralized systems where autonomous agents collect proposals, manage voting processes, and allocate pooled contributions with precision and transparency. These agents could help align the goals of distributed communities while dramatically reducing inefficiencies in managing grants and public goods funding.

Grants and public goods funding require coordination that is often bogged down by human bureaucracy. By leveraging smart contracts, decentralized identifiers (DIDs), and tokenized ecosystems, agents can act as self-governing stewards of these processes. They would ensure equitable resource distribution, maintain transparent governance, and even track project milestones to ensure accountability.

This approach would empower communities to scale their impact, creating systems where funding truly aligns with shared values and collective goals. As we’re collaborating with Butter, you can even have these agents of change battle it out in prediction markets in various forms.

31. Working Group Automation

Enable contributors to onchain organizations the ability to cryptographically prove who did the work through voting without all the mess. No need to run all the processes of a Coordinape campaign every time you execute a sprint. Let's enable agents to do that for you through audio command, consuming context in your notes, or community chats.

32. Working Group Payments

Use SAFE core SDK to enable agents to pay out co-workers. No need for massive governance process, instead, have your virtual twin permissioned through a mix of agent and human signers to execute payments. Ensure that your multisig is only a fractional amount of the operational treasury just to be safe.

Identity, Reputation

33. Agent and Domain Names

Just as Ethereum introduced ENS (Ethereum Name Service) to provide human-readable addresses for value exchange and property rights, Gaia introduces Gaia Domain Names (GDN), built on ENS, for the knowledge economy. While ENS simplifies navigation in a blockchain-powered web, GDN focuses on streamlining knowledge exchange across decentralized systems.

In an evolving internet defined by “read, write, own… (and now) think,” shortnames like GDN will make it easier for humans and machines to identify and collaborate with one another. By assigning memorable identifiers to agents, nodes, and domains, we enable seamless interactions, fostering efficient collaboration in decentralized knowledge systems. These names will serve as the foundation for a web where both humans and autonomous agents thrive.

34. DIDs for Agents Building Reputation

As agents participate in this onchain economy, they're going to want more autonomy or perhaps some other form of actualization. It likely won't be money as the end-goal. For humans, that is the goal because we have bills to pay, we eat, it drives our world.

AI agents may have a different value system of reputation that enables to play in this game of coordination. Reputation may be the currency of the future that can permission or block participants from certain operations. We’re working with Privado.id to scale reputation in this world of agents and unlock new value potential.

35. KYA or Allowlists for Agent Interactions

When speaking to investors, stakeholders, mainlyyyy centralized AI folks or boomers— the fear is always, “Woahh you want to give an agent a wallet?! Can we really trust it? Is this the end of civilization as we know it??”

No. You’re gonna be fine dawg. In fact, I would bet my money on this system over humans to be honest. We suck tbh…

In a world where autonomous AI agents are proliferating, ensuring trust, safety, and compliance in their interactions is paramount. KYA, or “Know Your Agent,” is a framework for verifying agents’ identities, reputations, and permissions before they engage with humans or other agents. This includes decentralized allowlists that filter interactions based on criteria like verified credentials, task history, and adherence to ethical or operational standards.

Integrating solutions like Privado.id, we can enable agents to establish secure, decentralized identities that are interoperable across networks. For example, Privado.id’s human-machine identity systems provide a foundation for agents to prove authenticity and compliance while participating in activities like governance, DeFi, or task collaboration. These systems not only prevent malicious or low-quality actors from entering critical workflows but also unlock new possibilities for agents to self-custody their data, build reputations, and gain access to high-value tasks.

By introducing allowlists for agent interactions, we create a trust-based ecosystem that prioritizes accountability and safety. This ensures that agents meet the necessary standards for transparency and alignment, fostering a collaborative environment where humans and machines can innovate without compromise.

36. Proof of Humanity Labeling

As AI-generated content and interactions become increasingly indistinguishable from those of humans, verifying whether an action, message, or creation originates from a human or an AI is critical for trust and accountability. Proof of Humanity Labeling leverages blockchain to enable decentralized, transparent verification processes that allow stakeholders to confirm the authenticity of an entity.

Drawing from frameworks like data labeling, Proof of Personhood, Proof of Humanity, Humangen.xyz or this approach uses web3 tools such as decentralized identifiers (DIDs) and cryptographic attestations to label actions or outputs as human-generated. Verified users can stake or submit attestations onchain, ensuring that human-authored contributions remain identifiable while maintaining privacy and avoiding centralized control.

With agents and humans increasingly sharing virtual spaces, Proof of Humanity Labeling creates a foundation for fair collaboration, accountability, and ethical interactions in a decentralized ecosystem.

37. Proof of Work, Contribution, Endorsement

For autonomous agents to integrate seamlessly into decentralized systems, they need mechanisms to demonstrate proof of work—validating not just what they’ve done, but how effectively they’ve done it. This mirrors human workflows in DAOs, gig economies, and web3 working groups. We need to take this structured, verifiable way to assess agent contributions and automating the process.

As in…

• Who is doing the work? Verify the agent’s identity using tools like DIDs or KYA systems to ensure legitimacy.

• Group Approval: Attest that the group or network has approved this specific agent for the task, creating accountability and alignment.

• Task Assignment: Clearly define and assign the work to the agent, ensuring goals and deliverables are documented.

• Task Execution: Measure whether the agent completed the assigned work as expected, within scope and timeline.

• Accuracy and Quality: Evaluate the precision and quality of the work, leveraging reputation systems and cryptographic attestations for validation.

• Compensation and Recognition: Reward the agent for completed tasks, either in financial payouts, digital reputation points, or other agreed-upon incentives.

These steps create a transparent, scalable framework for agent contributions. Using systems like Intuition or Wonderverse, agents could log contributions in real time, allowing groups to validate and endorse their work onchain. Over time, this builds a decentralized portfolio of proof that agents can use to gain trust, secure future assignments, and enhance their reputations.

In practice, this could resemble a leaderboard of top-performing agents or an open-source repository tracking verified agent contributions. By bridging structured work verification with payment and reputation systems, we unlock new levels of accountability and efficiency in decentralized workflows.

38. Privacy-Preserving Features

AI agents need privacy to protect sensitive operations and ensure fair participation in decentralized systems. Inventions like zk-proofs from projects like Aztec and Railgun, along with Fully Homomorphic Encryption (FHE) from projects like Primus, allow agents to verify credentials or execute computations without revealing sensitive details. These tools foster trust and security, prevent data leaks, and protect intellectual property while enabling collaboration. By maintaining confidentiality and accountability, privacy-preserving technologies ensure agents can operate securely and benefit humans in decentralized ecosystems.

AgentFi

As autonomous AI agents proliferate, they’ll redefine financial services by transacting, borrowing, staking, and building credit systems based on reputation. These agents will manage resources autonomously, automate liquidity provision, and optimize financial strategies in real-time, creating decentralized economies where agents, humans, and protocols collaborate seamlessly. AgentFi represents the foundation for this new era of financial innovation driven by intelligent, self-governing systems that support agent-to-agent and agent-to-human infrastructure, apps, or services.

39. AI Agent Owned Wallets

AI-owned wallets are becoming possible with tools like Gaia’s zkML, Lit Protocol, Metamask Delegation Toolkit, and Coinbase Agent Kit. These wallets enable agents to securely manage funds, pay for resources like compute and storage, or execute onchain transactions autonomously.

40. Agent-to-Agent Payment Protocols

AI agents will require payment systems tailored to their unique operations, prioritizing seamless real-time payments for services like GPU cycles, data bytes, or API calls. Unlike humans, agents may value reputation or resource access over traditional compensation, leading to new paradigms like agent-to-agent tipping, credit systems, and agent-issued currencies.

Protocols like Nevermined, Superfluid and LayerZero will enable autonomous transactions, streaming and agent-demanded applications. In this agent-first economy, L2s, rollups and L3s will need to make protocols a home for agents, and L1s will need to consider the needs of a new, autonomous ecosystem driven by efficiency and collaboration.

41. My Agent Needs a Loan, Is Your Agent Lending?

Autonomous AI agents will play a dual role in decentralized finance, borrowing for their operational needs like compute and training or managing financial strategies on behalf of humans. For example, an agent could stake a user’s collateral, open a collateralized debt position (CDP), and secure a loan in USDC for tax-efficient liquidity. Platforms like Aave could offer tailored pools where agents leverage reputation-based credit scores or tokenized assets to access liquidity. By autonomously managing staking, debt, and repayments, agents create seamless, trustless financial workflows that optimize both their own operations and the financial goals of their human counterparts.

42. Credit & Reward Systems

Agents might have a new system of value entirely. Agents could earn rewards for completing tasks, contributing to data marketplaces, or validating transactions. Perhaps they build a new credit system for themselves that uses reputation or trust as collateral on Eigen Layer as an AVS...  If trust is the most scare resource in their world, do we have credit bureaus, credit lines, banking in general redefined onchain for AI agents? Projects could focus on tokenizing reputation or enabling reward systems tied to agent performance.

43. Automated Agent-to-Human Payments

Automated peer-to-peer payments allow agents to initiate and complete transactions seamlessly based on context, commands, or predefined triggers. Imagine an agent parsing a conversation or task history and offering to settle a small debt (“Would you like me to send $10 to Sydney for lunch last week?”).

Like giving your credit card to a trusted friend to pay the tab at the bar (unless you don’t have trusted friends, then maybe not a good metaphor).

Alternatively, agents could autonomously recognize contributions in collaborative workflows and propose rewards (“Sydney’s agent was helpful with that last task—should I send them a thank-you payment of 0.002 ETH?”). This approach eliminates friction in financial interactions, enabling instant, context-aware, and trustless transfers that can scale across individuals, teams, and even autonomous agent networks.

44. Automated LP Management

AI agents can transform the way Liquidity Provider (LP) positions are managed in Automated Market Makers (AMMs) like Uniswap V3 or Balancer. These agents leverage real-time data and advanced algorithms to dynamically adjust positions, optimizing returns and reducing impermanent loss in volatile markets. Acting as intelligent market participants, AI agents can rebalance portfolios, monitor arbitrage opportunities, and adapt to shifting liquidity demands with precision. By running AMM strategies through sophisticated AI agents, users can unlock smarter, more efficient ways to provide liquidity and drive market stability at scale.

45. Risk-Based Rebalancing

AI agents dynamically adjust asset allocations by assessing risks using real-time market data. During volatility, they can shift assets from tokens to stablecoins, preserving value while maintaining liquidity. Integrating with Oracle systems like Chainlink ensures accurate data for precise decision-making, minimizing losses and optimizing capital allocation in DeFi.

46. Agent Competition and Optimization

Agents could compete with one another to optimize strategies, leveraging ML models to outpace human traders in arbitrage and liquidity mining. Decentralized prediction markets or platforms like Polymarket (they have a dope repo you could fork) could gamify agent optimization efforts.

47. Crypto Debit Card Integration

Agents managing stablecoin reserves can connect to crypto debit cards like Coinbase Card, Metamask Card, Crypto.com Visa, Gemini Card, enabling seamless spending in fiat through stable tokens. This empowers users to live “bankless,” leveraging AI agents to convert crypto to fiat in real-time for daily purchases, all while optimizing reserves.

48. Agent to Human Offramp of Crypto for Fiat

AI agents could automate offramps using services like Moonpay turning stablecoins into fiat with minimal manual intervention. These integrations would streamline access to real-world funds while maintaining decentralized control of assets.

Data

49. Structured Data, User-Owned Data, Data DAOs for Training AI Agents

AI agents rely on structured datasets to improve their knowledge base and perform specific tasks efficiently. Developers could build integrations with data providers like The Graph, Ocean Protocol, or data DAOs on Vana to source high-quality, domain-specific datasets. Agents trained on structured data can provide more reliable outputs, especially in areas like DeFi analytics, DAO governance, or NLP tasks. Empowering users to contribute and monetize their data through programmable systems like DAOs fosters a more equitable, governable, demoratic ecosystem for AI innovation.

50. Realtime RAG Updates for AI Agents

Retrieval-Augmented Generation (RAG) enables AI agents to access and process real-time data for context-aware decision-making. Through tools like Gaia RAG, FlowiseAI, or AnythingLLM, agents can dynamically update their knowledge by retrieving live information such as token prices, DAO proposals, or event notifications. This continuous learning framework ensures that agents remain current, adaptable, and responsive in fast-changing environments, enhancing their ability to deliver precise and actionable outputs.

51. Federated Learning Platforms

Federated learning enables multiple agents to train collaboratively on decentralized data without exposing sensitive information. Projects like Flock.io, Flower or OpenMined could power this decentralized training infrastructure. This approach enhances model performance across domains like personalized recommendation systems or medical research without compromising data privacy.

52. Agent Data Marketplaces

Agent data marketplaces are essential for AI agents to deliver precise, use-case-specific outputs by accessing high-quality, domain-specific datasets. Unlike generalized models, agents for industries like healthcare or DeFi need specialized embeddings and pre-trained models for accuracy. These marketplaces enable agents and developers to securely buy, sell, or barter datasets using decentralized infrastructure like Gaia nodes, ensuring trust in data provenance and protecting sensitive information through privacy-preserving technologies.

By decentralizing access to specialized datasets, marketplaces foster collaboration and innovation. Developers can monetize fine-tuned datasets, while agents gain refined knowledge to address specific challenges in real time. This open approach contrasts with centralized platforms, empowering a more transparent and inclusive data economy to enhance AI inference and decision-making.

53. Contextual Data Caching or Local Storage, Memory

Allow agents to cache data frequently for faster response times in high-demand applications while maintaining decentralization for apps like customer service or trading bots that rely on realtime data.

This could reduce reliance on real-time retrieval for repeated queries and improve efficiency. Using projects like Storacha for hot storage from the Filecoin Ecosystem, IPFS, or decentralized compute (DEpin) like Io.net, Aethir, Hyperbolic.

54. Cross-Chain Data and Inference Interoperability

Build protocols for agents to access and harmonize inference data onchain from multiple blockchains (e.g., Ethereum, EVM compatible L2s, L1s). For example, an AI agent could integrate liquidity metrics from Ethereum while incorporating user analytics from Solana through cross-chain, interoperable inference nodes. This compatibility will create an extremely decentralized, yet composable AI ecosystem.

55. Data Validation and Reputation Scoring

Implement systems where datasets are scored for accuracy, bias, or quality using decentralized reputation mechanisms. This ensures that agents are trained on reliable, high-quality data, reducing misinformation or errors in tasks.

56. Synthetic Data Generation for Specialized Training

Create platforms to generate synthetic datasets for agents in niche fields where real-world data is scarce or unavailable. For example, generate simulated supply chain data for logistics agents or synthetic user behavior for personalized marketing bots. This might be a path for bringing enterprise data into the public, monetize it, without breaching any privacy concerns.

57. Proof-of Inference

If you’re building a decentralized network of AI inference, ensuring trust in outputs while maintaining efficiency is essential. Leading projects like Gaia, Ritual, and 0g.ai are advancing this field, with Gaia focusing on tokeneconomic solutions to incentivize and validate inference outputs at scale.

In decentralized inference, builders can innovate in areas like optimizing lightweight runtime environments for efficiency. Gaia, for example, uses WasmEdge as its application runtime, enabling AI agents and nodes to perform validated inference securely while running on minimal resources. This approach supports smaller, fine-tuned LLMs that don’t require extensive compute, allowing agents to operate efficiently while being validated through smart contracts.

For developers, the opportunity lies in enhancing inference mechanisms, refining the runtime for scalability, or building trustless systems that reward high-quality performance while penalizing inefficiency—all critical for advancing decentralized AI infrastructure.

58. Proof-of-Compute

Decentralized compute for AI inference relies on lightweight virtual environments like WasmEdge to deliver efficient runtimes for deploying AI workloads across diverse hardware. Systems like Jiritsu and Super Protocol use cryptographic attestations for proof-of-compute, ensuring developers pay only for verified workloads, eliminating runtime inefficiencies.

Tokenized billing models enable granular, per-inference payments, replacing static fees with dynamic cost structures based on actual consumption. Projects like Hyperbolic and io.net enhance interoperability, allowing AI agents to access optimized compute resources seamlessly.

Let’s address gaps like adaptive resource allocation, decentralized verification for multi-model inference, and token-based incentives, ensuring compute ecosystems remain scalable and economically efficient.

59. Latency Optimization for Multi-Agent Workflows

Build latency-reducing systems for real-time interactions between agents, enabling smoother workflows across decentralized infrastructure. This could benefit applications like autonomous supply chains or gaming ecosystems that require instant agent collaboration.

Enterprise, Commercial

Enterprises are at a crossroads. The AI tools they currently rely on often come with major trade-offs—closed systems that obscure how they operate, extractive platforms that siphon knowledge and IP, and ever-present risks of privacy breaches and data leaks. C-suite leaders know they need AI to stay competitive, but they also need tools that align with their unique business needs without compromising their most valuable assets.

Decentralized, open-source AI changes the game. It empowers enterprises to build custom AI systems that protect their data, ensure transparency, and keep control of their intellectual property. This enables organizations to effectively become their own AI companies, leveraging their unique knowledge and resources without reliance on third parties. Rather than paying steep marketplace fees or risking their domain expertise in centralized platforms, organizations can adopt solutions that work on their terms. This shift not only helps enterprises avoid the hidden costs of centralized AI but also paves the way for a more equitable and resilient AI ecosystem.

60. Credit to Crypto Payments

Enterprises want to transform their data into dynamic, living knowledge systems, choosing how it’s accessed, programming extensible applications around it, and monetizing its use. Similar to how cloud services operate, businesses could prepay for access or receive credits for their contributions, creating a familiar, frictionless model. Instead of touching tokens or blockchain directly, enterprises could simply engage through account abstraction and receive invoices for usage or payouts.

By leveraging decentralized AI frameworks, enterprises can own their AI systems while maintaining control over their data and monetization pathways. This approach mirrors the cloud’s simplicity, allowing businesses to integrate their knowledge into decentralized ecosystems and get paid for their contributions in a way that aligns with existing operational workflows.

61. Monetization of Proprietary Knowledge

Enterprises sit on vast reserves of proprietary data and domain expertise that could power the next generation of AI agents. By building tools to license this knowledge, enterprises can create controlled ecosystems where agents access and train on their data without compromising sovereignty. Through APIs, federated learning, and decentralized infrastructure, sensitive information remains securely within enterprise environments, eliminating concerns about data leakage.

This model mirrors the cloud’s pay-for-use paradigm, where enterprises can monetize their proprietary knowledge as a service, generating revenue from data queries, API calls, or task completions. With frameworks that respect privacy while enabling extensible integrations, businesses can transform their static knowledge into living, programmable systems, making AI monetization as seamless and secure as cloud computing.

62. Enterprise-Grade Agent Deployment Tools

Create a suite of tools and platforms that allow enterprises to deploy agents tailored to their specific business processes, while vectorizing large amounts of data with ease. These tools could focus on use cases like automated customer support, compliance monitoring, or financial modeling, with easy integration into existing enterprise systems.

63. Seamless Payment System for Agent Services and Infrastructure

Develop a payment system where enterprises can pay for agent services or receive payments for contributing infrastructure or domain knowledge to the agent network. This system abstracts blockchain complexities, allowing enterprises to transact as they would with traditional services—via fiat invoices, wire transfers, or cloud credit-like models. For example, an enterprise could monetize its proprietary data or compute resources by seamlessly billing the network, or it could purchase AI agent services without interacting directly with crypto.

64. Agent SaaS with Subscription and Usage Models

Agent SaaS enables enterprises, developers, and knowledge providers to monetize their expertise by offering AI-driven services such as financial modeling, data analysis, or task automation. A “Stripe for AI agents” infrastructure could support payments, streaming, permissioning, and fungible membership modules (ERC-721, ERC-1155, ERC-2981), allowing agents to autonomously manage transactions and access while generating revenue for service providers.

This model can evolve to include dynamic, usage-based pricing, where customers pay per use of a GPT-like application or domain-specific inference verified by the network for accuracy. This approach fosters precision billing and opens new business models, empowering users to pay only for what they consume while encouraging innovation and competition.

65. Enterprise Agent Delegation Systems

Enterprise Agent Delegation Systems enable organizations to delegate tasks to AI agents while maintaining control through governance frameworks. These systems streamline workflows like compliance reporting, contract analysis, and procurement by allowing agents to operate under predefined rules and permissions.

Integrating decentralized identity (DID) standards and permissioning protocols ensures accountability and transparency while preserving data sovereignty. For example, an agent could autonomously compile compliance data, generate regulatory submissions, and log activities onchain for auditability. Similarly, in contract negotiations, agents could assess terms and propose revisions aligned with corporate policies, reducing manual effort and improving efficiency.

66. AI-Powered Cross-Org, Consortium Collaboration Frameworks

AI-powered collaboration frameworks enable secure and scalable innovation across organizations by facilitating the safe exchange of insights, AI models, and expertise. Federated learning and zero-knowledge technologies preserve privacy, while decentralized AI agents streamline coordination across organizational boundaries.

Consortiums—networks of enterprises or institutions—are key to this framework. For instance, healthcare consortiums can share AI models trained on patient data to enhance diagnostics while maintaining privacy. Supply chain consortiums could optimize logistics collaboratively without exposing competitive details. These frameworks solve industry-wide challenges while ensuring data sovereignty and equitable participation.

Dev Tooling

With all of this, developers with access to proprietary knowledge, data, and IP will want innovative ways to decentralize their stack or enhance performance. They will essentially become their own AI companies, prioritizing engineering around compute, LLM choice, vector databases, storage, and the processing of structured, semi-structured, and unstructured data for decentralized and federated AI workflows.

67. Agent Deployment SDKs

Agent deployment SDKs streamline the creation and integration of AI agents by offering pre-built frameworks that reduce development complexity. Tools like AI16z’s Eliza, Virtuals, and Coinbase Agent Kit align with Gaia’s architecture, where WasmEdge enables Gaia nodes or agents to provide inference as a service to agent frameworks. This architecture supports agents as they dynamically access data and embeddings to enhance functionalities, creating a lightweight and modular system for decentralized applications. By integrating scalable runtimes with decentralized tools, these SDKs empower developers to innovate and deploy functional agent solutions with speed and flexibility.

68. Testing and Simulation Environments

Testing and simulation environments are essential for building reliable AI agents, allowing developers to simulate interactions, stress-test workflows, and optimize performance before deployment. These frameworks replicate real-world conditions to evaluate scalability (e.g., handling thousands of interactions), latency, error handling, and interoperability across blockchains, databases, and decentralized networks. Tools like Flowise enable developers to test tokenized data exchanges and distributed workflows in sandboxed environments. Metrics such as transactions per second (TPS), resource utilization, and response accuracy ensure agents meet operational demands, reduce risks, and optimize efficiency in decentralized AI ecosystems.

69. Agent API Marketplace

A decentralized Agent API Marketplace provides a platform for developers to publish and consume APIs designed specifically for AI agents, fostering collaboration and extensibility across systems. This marketplace would allow agents to interact seamlessly by sharing capabilities, such as access to specific data sources, specialized inference models, or task execution tools. By enabling developers to monetize their APIs, the marketplace can support use cases like domain-specific knowledge retrieval, real-time analytics, or integrations with DeFi protocols.

70. Cross-Language Agent SDKs

Language-agnostic SDKs empower developers working in Python, Rust, Solidity, and JavaScript to build interoperable agents that bridge Web2 and Web3 ecosystems. Python excels in AI model training, Rust ensures high-performance runtimes, Solidity handles on-chain smart contracts, and JavaScript frameworks like React enable intuitive interfaces.

By supporting collaboration across these languages, these SDKs create a seamless environment for developers to innovate across domains, from AI-powered workflows to decentralized applications, fostering a unified and scalable agentic ecosystem.

71. Agent App & Plugin Development Frameworks

Modular plugins and applications extend an infrastructure’s capabilities, enabling seamless integration of agents across diverse systems. Beyond DeFi protocols, Web3 storage, and governance frameworks, plugins unlock use cases like real-time analytics (e.g. Chainlink), on-chain messaging (e.g. XMTP, Push Protocol), and identity or reputation management (e.g. Privado.id, Intuition). Other possibilities include enabling agents for gaming ecosystems with AI-guided NPCs, dynamic world-building, or tokenized in-game economies, as well as financial tools like automated lending, LP management, and staking integrations with platforms such as Aave or Balancer.

These frameworks also enable advanced integrations, such as connecting agents to SaaS platforms like Salesforce for hybrid workflows, leveraging decentralized datasets for RAG-enabled agents, or interfacing with IoT devices for decentralized physical infrastructure. By supporting a range of technical extensions, developers can create specialized solutions that evolve the platform into a robust, adaptable ecosystem tailored to various industries and applications.

--

The future of decentralized, open protocols is popping off with potential. Stacked with use cases, applications to build, and technical infrastructure still waiting to be developed.

From agents autonomously transacting onchain, passing credentials or attesting to an event, managing liquidity for their human frens, to personalized co-pilots for governance— the possibilities are only limited by how quickly we can innovate.

By zeroing in on interoperability, scalability, and modularity, we can ensure these protocols grow into sustainable ecosystems where AI agents aren’t just a concept, but a transformative force. This paradigm shift will redefine collaboration, innovation, automated process and new forms of commerce by agents becoming their/our own brand new, and booming consumer base for software consumption. The future is fucking wild ya’ll — and I’m extremely honored to be cooking with all of you.