The AI boom is in full swing, but crypto VCs are struggling to get a seat at the top AI project tables. Instead, they settle for second best, investing in “borderline AI” projects within their Crypto comfort zone. This article examines five recent on-paper AI projects that have completed funding, revealing how crypto capital uses the “AI” label to exchange for liquidity.
(Background: The watershed moment for crypto VCs: the survival answers of a16z, Dragonfly, and Paradigm)
(Additional context: Crypto VCs are telling the story of $2 trillion on Wall Street)
Table of Contents
Toggle
- Derivio: A DEX that doesn’t do Meme tools isn’t a good AI trading platform
- Superpower: The vanished AI agent product
- Finrob: Proprietary models in crypto that can’t be used
- PlutonAI: The DeFAI hiding in the corner
- Unicity: The real master of big-picture dreaming
- Are crypto VCs fools?
The popularity of AI has thrown Web3-focused VCs into confusion. On February 28, Paradigm announced plans to raise a new fund focused on AI and robotics, with a target size of up to $1.5 billion. I analyzed this as a signal that crypto capital is shifting its focus from Web3 to AI industries. When there are no good projects to invest in within crypto, the booming AI sector becomes a new battlefield for crypto capital.
But I overlooked one thing: top AI projects are rarely accessible to crypto VCs. Good projects tend to prioritize investors with experience and resources. Paradigm, with its reputation, might still squeeze into the top-tier AI investment circle, but small crypto VCs with no reputation in traditional finance can only watch from the sidelines as the big players eat the meat.
Is it hopeless? Of course not. There are always ways. If you can’t invest in top-tier AI projects, then settle for projects tangentially related to AI within the crypto comfort zone—it’s still a ticket to ride the wave of the era.
As internet memes joke, if a Web3 company just changes all “loading…” to “thinking…”, it can rebrand as an AI startup. Under crypto VCs’ anxiety, some Crypto+AI projects can secure millions of dollars in funding with just a white paper and a product without PMF.
To illustrate this industry abstraction, I selected five recent “on-paper AI projects” that completed funding rounds.
Derivio: A DEX that doesn’t do Meme tools isn’t a good AI trading platform
A fresh example is Derivio, an AI-native trading platform announced on March 18 that completed $6 million in funding, with investors including YZiLabs and other crypto VCs. However, the official statement clarified that the $6 million is the total amount raised so far, not just from this round.
It’s rare to see early-stage projects announce accumulated funding without specifying the new raise—do VCs feel embarrassed about the small amount? Or is it a PR move to align with AI transformation? According to research, Derivio was a Binance Labs (now YZiLabs) incubated project in Q2 2023, back when it was a decentralized derivatives trading platform on zkSync.
In 2024, Derivio launched the Ethereum Layer 2 Derivio Network, claiming full compatibility with Solana Virtual Machine (SVM), trying to develop both ecosystems. Unfortunately, it failed.
Now, if you visit Derivio’s website, instead of trading charts or typical DEX swap pages, you see Pump.fun’s chain monitoring dashboard.
Yes, you read that right. This former decentralized derivatives platform, now an AI-native trading platform, mainly scans chains for Pump traders. A DEX that doesn’t do Meme tools isn’t a good AI trading platform. Moreover, you can even buy tokenized US stocks on this platform.
As an AI-native trading platform, Derivio should have AI features, right? But clicking on “Agent” in the top left shows a “COMING SOON” page…
Although Derivio currently has nothing, its vision is imaginative. In an article on X titled “The Last Generation of Human Traders,” they wrote that most trading terminals are designed for humans. Derivio aims to build the first full-stack trading terminal designed for AI, developing an efficient data stream engine from scratch to minimize latency from on-chain events to AI processing.
I found this idea fascinating—something only an advanced AI could conceive. But a quick check revealed the article wasn’t written by a person but generated by AI. Derivio is ahead of the curve—before replacing human traders, it replaced its own staff with AI.
Superpower: The vanished AI agent product
Superpower is an AI agent revenue protocol. On March 6, it announced a pre-seed funding round, with investors including Taisu Ventures, Paper Ventures, CatcherVC, and 280 Capital, though the amount was undisclosed. Superpower aims to build a platform enabling AI agents to generate income, access funding channels, and realize capital appreciation.
Achieving such a grand vision requires accumulated technology and ongoing development, but Superpower doesn’t even have a website yet. March 6 was not only their funding announcement day but also the day they posted their first tweet.
Unlike Derivio, which has no AI Agent but at least some interaction, Superpower’s website is a visual shock. It has no clickable buttons—only a slideshow with the phrase “YOUR AGENT IS BROKE AF.” I don’t even understand what that means; the slides move too fast, and I can’t tell if the last word is “AI” or “AF.”
On Superpower’s official X account, I found a link to a project called Prolly, a prediction market platform, with a post explaining how agents can make money on Prolly. I wanted to try, but the product requires an invite code, so I gave up.
Of course, all posts introducing the project on Superpower’s homepage are also AI-written.
Finrob: Proprietary models in crypto that can’t be used
Finrob is an AI-driven crypto market research platform. On February 25, it completed a $3.9 million seed round, with investors including Maven11, Placeholder, Archetype, Fabric Ventures, Dispersion Capital, and Node Capital. What does this project do?
In simple terms, it’s a conversational large model, similar to ChatGPT or Gemini, and Finrob integrates with these models.
The question is: what’s the difference between chatting with ChatGPT or Gemini directly and using Finrob? Finrob claims it’s tailored for crypto, with real-time data integration, on-chain analysis, and specialized tools.
Specifically, it connects with CoinGecko for real-time prices and market data, Glassnode for over 200 on-chain analytics, Tavily for web searches and news, Perplexity for in-depth research, and other sources like DefiLlama, Etherscan, and LunarCrush for social sentiment.
In other words, once Finrob connects these data sources to a free large model (it doesn’t support high-end models like ChatGPT 5.4), it can claim to be “built for crypto.” At $3.9 million, it’s cheaper than training a large model from scratch.
From their use cases, it seems Finrob’s ultimate goal is to provide investment advice. But can AI truly guide real trading decisions? Will Finrob’s free models outperform GPT 5.4 or Claude Opus 4.6? I initially thought Finrob would be better at real-time token prices, but I tested it and found that ChatGPT 5.4 can also fetch BTC prices from CoinGecko just fine.
Despite my skepticism, I still wanted to try Finrob myself, but it kept throwing errors regardless of whether I logged in with email or wallet.
PlutonAI: The DeFAI hiding in the corner
PlutonAI is a DeFAI platform aiming to let AI agents analyze markets, optimize strategies, manage yield opportunities, and perform complex on-chain operations. On February 17, it completed a $2.7 million private funding round led by KitchenVC, with HyperGPT participating.
From its positioning, PlutonAI looks like a typical Crypto+AI project. Whether DeFAI has product-market fit, safety, or can help users make money is another question. Today, with rapid AI tech iteration, letting AI replace humans in on-chain operations is quite straightforward.
Especially after OpenClaw’s explosion, the crypto world has seen a lobster farming craze, with many influencers sharing how to install OpenClaw and use it for on-chain trading or market prediction. Meanwhile, platforms like Binance and OKX have launched AI agent trading tools or assistive features.
Thus, even DeFAI can be considered a pseudo-concept.
I also wanted to try out PlutonAI’s AI Agent, but I still couldn’t log in.
Unicity: The true master of big-picture dreaming
Unicity is an infrastructure developer aiming to build the “Agentic Autonomous Internet,” enabling billions of AI Agents to discover, trade, and settle trustlessly at machine speed. On February 19, it completed a $3 million seed round, with Blockchange leading, and Outlier Ventures and Tawasal participating.
This is another highly ambitious company. To understand what they’re really doing, I read their white paper (finally not AI-written). Overall, Unicity believes current blockchains are inadequate in throughput, latency, privacy, and cost for high-frequency AI trading and collaboration. They plan to create a bottom-layer network for AI Agents, moving all transactions off-chain in a peer-to-peer manner, with only state changes recorded on-chain, preventing double-spending, and using verifiable Agents to supervise execution.
Compared to the previous projects, Unicity sounds very serious and might even develop into a real public chain ecosystem. But since it was founded in 2025, what has it actually delivered? No testnet, no ecosystem—its March 10 update mentions plans to launch an agent operating system called AstridOS, allowing tools like Claude Code and OpenClaw to run on it. But whether anyone will actually use it remains to be seen.
Are crypto VCs fools?
Of course, among the recent AI funding rounds, some projects are solid.
For example, RoboForce, which closed $52 million on March 17 led by YZi Labs. It’s an AI robot company with no direct crypto ties. The project is building physical AI robots, and NVIDIA CEO Jensen Huang mentioned this sector as a future direction at GTC 2026. RoboForce’s robots also appeared at NVIDIA GTC 2025.
Another example: Kled, which raised $5.5 million on March 11, focusing on AI data markets; and VeryAI, which raised $10 million on March 12, building AI Agent identity systems. These projects have product-market fit, real products, and performance.
But such projects are rare in crypto. So, are other crypto VCs fools? Knowing many so-called AI projects lack real product logic, why do they still pour money in?
The answer lies in the surface—the liquidity of the “AI” label.
Although they can’t get into top-tier project tables, money still needs to be spent. Creating “AI dummies” within crypto is a business opportunity. Projects only need to package themselves as AI, and VCs will blindly invest. As long as both sides can eventually recoup their investments, it’s considered a good AI.
