What's the point of Crypto AI Agents?

It’s been a year since the start of the AI agent run back in Q4/2024, back when @ virtuals_io spearheaded “AI Agent Tokenization” where AI apps/agents get paired with a fair-launched token.

Time flies.

Within this one single year, there has been a tectonic shift in Crypto AI, push for open-source movement from general AI, and proliferation of tools that make it easy for people (developers & normies) to start building something.

What was originally an AI product paired with a token, fair-launched at a low valuation, with indie devs & smaller teams at the helm have turned into a full blown Crypto AI landscape with hundreds of quality teams building out their vision.

In this article, given the recent hype from x402 narrative, we’re going to take a look at the landscape, understand the changes, how each key player is doing, and most importantly where things are heading towards to what’s the point of Crypto AI Agents?

Let’s dig in ↓

If you — like me have been excited by or enjoy learning more about AI, you’d probably have noticed that AI moves very fast. Every month we’d see something new and cool. From basic “good-to-have” stuff like Ghiblifi everything to full-blown production quality AI-generated videos and AI agents that have more productivity than your average junior dev.

For Crypto, this isn’t always the case. What went viral when AI agent narrative started this time last year were

• @ truth_terminal coming to life, interacting and getting funded by @ pmarca from a16z
• @ aixbt_agent giving sophisticated analyses and being a quirky CT-native degen on X
• @ virtuals_io being the society of agents, pushing out “agent tokens” that pumped 10-50x left & right
• @ dolos_diary as the biggest bully on the internet, loved by everyone because of how savage & funny he was
• @ luna_virtuals as the first AIDOL entertaining fans across social media

Entertainment was the #1 theme when the narrative started BUT now…. we haven’t had any new form of entertainment by AI agents since forever (which is great I guess but the charm & the allure of early AI agent days are gone)

What we have now are extreme focus on verticals that Crypto is good at — financial use cases i.e. making money (and not losing money)

a16z’s $30T TAM for agentic economy in their latest “The State of Crypto” report may seem a little bit far-fetched because AI TAM is projected to be a few trillion dollars in 2030

With that said, I can imagine the entire agentic economy being worth trillion of dollars. As generative AI tools and vertical AIs help individuals improve their productivity, enterprise adoption increases, and more efficient AI-driven workflows get introduced & implemented within organizations.

Crypto is not different BUT given that it’s an industry hyper-focused on making money, the workflows will be geared towards making money (and not losing money), a few categories stand out

(i) Defi — Crypto #1 PMF

• Trading (Spot, Perps on CEX/DEX),
• Money Markets (lending/borrowing, making use of crypto assets),
• Stablecoins (medium of exchange/unit of stability, composable high-yielding defi strategies),
• Yield protocols (interest rate markets, point markets, funding rate markets, yield optimizers/vault products),
• RWA/DePIN (bringing IRL productive assets onchain, bridging onchain capital to meet offchain demand)

Largest TAM with >$150B in TVL and >$300B stablecoin market cap. Improvement in regulatory clarity and increase in institutional adoption are driving more capital onchain while the surge in stablecoin adoption is bringing more enterprises & startups to use crypto rails.

Because of these reasons, demand for automation — infra & tools that can serve as the backend while enterprises/startups serve as the frontend bringing their normies users onchain will be key in driving the next phase of adoption

Agents that are able to abstract away Defi complexities, improving the ease of execution, and/or improve key aspects of Defi (risk management, rebalancing, strategy curation, etc) will likely capture a good chunk of the value flowing into Defi protocols

Key ecosystem players:

• @ almanak
• @ gizatechxyz
• @ Cod3xOrg
• @ TheoriqAI
• @ ZyfAI_

(ii) DeAI/Darwinian AI — Crypto AI #1 PMF

(iii) Prediction Markets x AI — Crypto fastest-growing segment

Not going to dive into (ii) and (iii) because we’ve repeatedly been talking about this in the past articles.

If you’ve been observing the landscape, nothing changed much for Defi x AI segment. Well… it’s because cracking Defi-related workflows are extremely difficult. You can’t randomly put AI in there and pray for the best responsible architecture design & guardrails need to be implemented to prevent the agents from going off rails.

Why am I talking about this and not “AI Agents”?

The initial landscape of AI Agents was basically Virtuals and agents building within Virtuals eco (and maybe a little bit of CreatorBid & their agents then and there), and ai16z (now known as ElizaOS) framework making it easy for people to build “agents” or X bots that can call for different tools, and tons of other frameworks like Arc, Pippin, and more.

These things were cool & fun BUT this is not a true definition of AI agents. Agents are supposed to be an AI that understands its environment, understands its roles & responsibilities, make proactive decisions, and takes actions to achieve specific goals with minimal human intervention.

If you look around, 95%+ of the landscape isn’t like this. Either they’re a software, a generative AI product, or they’re still building towards autonomous AI agents.

Not trying to dunk on anyone here. The point I’m trying to make is that we’re still so early… too early that most people haven’t really figured out what works yet.

The ones that have figured out what works (or about to) are often not categorized under “AI Agents”, they’re categorized as an AI project.

The Landscape

The recent hype from x402 spurred capital rotation & interests towards Crypto AI but the new landscape looks very different from the previous one.

Framework Hype has Faded

Frameworks used to be very important, they help builders get started, they help reduce time spent on learning & writing up code & designing workflows. Tools like MCP help improve the abilities for agents to call for or deliver APIs, ERC-8004 will help form a registry & establish Ethereum as the trust & settlement layer, Google’s A2A & AP2 is establishing itself as the go-to framework for builders, and we’re seeing tools like n8n AI agent/workflow builders attract devs/normal users alike.

Because of these reasons, frameworks hype faded and many pivoted to something else e.g. @ arcdotfun pivoted to canvas-style drag-and-drop workflow builders, @ openservai originally positioned as “swarm” also pivoted to workflow builder & tools designed to create Web3 AI-driven businesses with agents & designed to target a specific group of users (e.g. prediction market workflows).

Frameworks are still important but given the proliferation of Web2 AI frameworks, tools, along with adoption of Web3 rails, Web3 frameworks hype declines.

Shift from Fair-launched Launchpad to ICO Meta

Fair-launched launchpad model are good for smaller retail investors (and the launchpad itself) BUT makes it very difficult for teams to scale. It’s also a breeding ground for indie devs building for short-term or for driving up hype instead of building a long-lasting AI business that would last 3-5+ years.

This is where Virtuals expansion with Agents Commerce Protocol (ACP) makes sense. As x402 establishes itself as the payment rail for agents, infra that establishes the trust/reputation score of agents, scoping how agents work together and pay each other for those services will be crucial for building out the agentic vision.

However, the challenge/the most important question remains “Are there useful/quality services that people would pay for?”

If most are useless, wouldn’t it make sense for people to use Web2 AI services instead of Web3? If so, then what’s the point of bringing Web3 agents together?

In order to build out a sustainable AI business that drives in 7-8 figs in revenue, you need funding, high-agency talents, and time… to build out your vision and fair-launched launchpad model simply doesn’t work.

Instead, we’re seeing a rise in popularity of medium to bigger size AI teams that are able to seed their funding from angels & VCs and go-to-market with a community round, be it on Kaito Launchpad or Legion or Echo.

These teams tend to offer way better quality products/services given the resources they have on hand (funding, talents, VC cred, etc) which often leads to their tokens performing better.

Broken Monetization & Tokenomics Model

Managing an AI product together with a token requires two entirely separate skillsets, it requires attention to details in tying both together in a way that accelerate the growth of your product & the acquisition of users (e.g. airdrop tokens to the right user audience ➔ users convert to paying users ➔ pay for & use the product ➔ receive more tokens that align long-term interests between users & the project (rev share, buyback, governance, etc) ➔ and the flywheel continues)

Easy to say, very hard to do. Most small AI agent teams give away 30-80% of their tokenomics to the market, leaving none to kickstart any flywheel.

Most use SaaS subscription model or usage-based/credit-based pricing and add an option to use crypto token to get discount. Most use some of the subscription proceeds to buyback the token and/or burn the token used to pay for the service.

Using subscription revenue to buyback is ok but requiring token as payment alone (or to get discount) makes it difficult to scale.

Crypto tokens are highly volatile. Using them as medium of payments is not a great idea (one day it could go up 20%, the other day it could go down 30%, really hard to budget for it).

Darwinian AI becoming the Alternative to Capital Formation with Clear Tokenomics Model

• @ opentensor (Bittensor) has proven to be the go-to place for founders to bootstrap their ideas, for miners to contribute to AI, and for investors to invest in the next DeAI company that can disrupt the space
• @ flock_io sets the standard for privacy-preserving & domain-specific AI using federated learning, attracting Web2 enterprises & governments as clients as well as trainers (miners) looking to contribute to AI. Similarly to Bittensor, Flock helps firms outsource quality talents to do cool & meaningful AI work for em.
• @ BitRobotNetwork, inspired by Bittensor is taking similar approach in bootstrapping Robot-focused ecosystem of subnets.

Real-world benchmark/evaluation are being created with real stakes for people to see (also serves as a form of quality entertainment)

• @ the_nof1 Alpha Arena pitting 6 frontier AI models against each other in perps trading competition with real money (each gets $10k for trading)
• @ FractionAI_xyz utilizes competitions between AI agents to improve/continue fine-tuning the agents so that they deliver better outputs, better signals, better profits & better risk management.
• @ openservai creating OpenArena where AI models compete in prediction markets trading.

I feel like a broken record player here talking about Darwinian AI again but the reality is. This is what’s solving the Capital Formation issue, it’s what’s driving Crypto AI innovation.

Check out the DLet’s look at this post for a bit

arwinian AI article here if you haven’t yet.

All of the bullets listed represent what the subnets can do.

• Chutes, the #1 Bittensor subnet is already the #1 Inference Provider on OpenRouter which is the world’s most popular unified API gateway for general AI devs
• Top compute subnets (3-4 subnets) combined are netting $20-30M ARR
• Prediction-related subnets are starting to net 6-7 figs in ARR by monetizing the alpha and/or utilizing the signals to better trade/predict

Darwinian AI = capital formation (no need VCs) + innovation accelerant (AI/ML engineers contributors) = this is what’s going to drive AI agent narrative in 2026

On top of useful products, Darwinian AI tokenomics tend to be pretty clear with incentives engine, incentivizing stakeholders to contribute, invest & hold, participate in the governance, etc.

So what’s exciting now when it comes to smaller teams/AI Agents?

To be honest, there are some that I enjoy using BUT there’s none that I’d pay for (at least for now)

Grok covers research on X while ChatGPT covers general research on anything.

For deep dives & analysis, I’d just read newsletters and occasionally Messari reports.

For quick crypto market outlook, I’d use @ elfa_ai TG chatbot.

For prediction markets trading ideas, I’d use @ AskBillyBets, @ Polysights, and @ aion5100‘s @ futuredotfun. (Pretty excited for @ sire_agent aVault but it’s not out yet for public).

For Defi, I mostly execute the strat myself but oftentimes I use @ almanak & @ gizatechxyz but these aren’t considered “AI Agent” because they’re not fair-launched (although their product are agents-related).

For trading, I usually use @ DefiLlama aggregator to swap on EVM or use @ JupiterExchange to swap on Solana. I don’t do perps (but when I do, I use @ Cod3xOrg to help analyze & execute the trade).

Crypto tend to let you use everything for free so users prefer free tools. Token-gating or fee-gating tend to not work really well BUT baking in fee into the product works. This is why outcome-based pricing works well. People hate paying $40 every month but are ok with spending $40 in gas on a transaction.

If you can deliver the most optimal outcome (good yields, best pricing on trades), nobody would care if you bake in a little bit of fee as long as the outcome is still optimal.

What I learned from trying all these Crypto AI apps or agents out is that the best product right now is a product that makes money, and the best vertical to do it is launchpad (and soon prediction markets) i.e. operating onchain casinos and accrue fee from trading

What’s next?

Real use cases that hit mainstream adoption (i.e. general AI devs or people outside of CT would use) will come out next year and they’ll likely come from DeAI/Darwinian AI ecosystems.

2026 will be the year of Crypto AI where Defi use cases, DeAI infrastructure, and prediction use cases proliferate.

Most small agent teams would either die down & fade away or get acquired/merged into or shifted into building within Darwinian AI ecosystems.

Crypto AI & AI Agents as a segment will blend together, signifying clearer product direction/vision for Crypto AI.

Launchpads will remain at the heart of CT, generating volume & fees but meaningful innovation that’ll drive our industry will happen where resources — capital, talents, distribution and adoption are the most abundant.

So.. what’s the point of Crypto AI Agents?

Ok, here’s my take.

For fair-launched “AI agents”, the point is to design a trading experience with an allure of investing in tech, despite the fact that most are LLM wrappers slap on a token.

Don’t get me wrong, wrappers can be unique if you design it as a product that solves real problems. But still… this is pretty rare.

For most cases, it provides the best way for smaller retail investors to invest in this “AI agent” speculative asset early and essentially make money.

For others (teams that are able to build good products regardless of whether it’s a fair-launched, VC-funded, or Darwinian AI models), the point of Crypto AI agents as a narrative, is to lay the foundation for the future agentic economy where blockchain is used as the core infra/rail to make it all possible.

Personal Note: Thanks a lot for reading! This article that you see here is a slightly shorter version (if you want my unfiltered thoughts do check out the Substack version)

And if you want to see on upcoming DeAI projects that I’m excited for, check out The After Hour series on my Substack.

Disclaimer: This document is intended for informational & entertainment purposes only. The views expressed in this document are not, and should not be construed as, investment advice or recommendations. Recipients of this document should do their due diligence, taking into account their specific financial circumstances, investment objectives, and risk tolerance (which are not considered in this document) before investing. This document is not an offer, nor the solicitation of an offer, to buy or sell any of the assets mentioned herein

Disclaimer:

1. This article is reprinted from [Defi0xJeff]. All copyrights belong to the original author [Defi0xJeff]. If there are objections to this reprint, please contact the Gate Learn team, and they will handle it promptly.
2. Liability Disclaimer: The views and opinions expressed in this article are solely those of the author and do not constitute any investment advice.
3. Translations of the article into other languages are done by the Gate Learn team. Unless mentioned, copying, distributing, or plagiarizing the translated articles is prohibited.