Web3 + AI at ETHDenver 2025
What better way to pulse check on Web3 + AI in 2025 than exploring the standout talks from ETHDenver?
Hello, everyone! Long time no see 😊
This is the Web3 + AI newsletter, a space dedicated to the various use cases, products, and services born in the intersection of blockchain and artificial intelligence. I hadn’t published in a long while, but I very much appreciate you still being here.
The field of Decentralized AI is so multifaceted and dynamic that in the few months I was away, it transformed beyond recognition. Or probably ‘evolved’ is the right verb here: after spending a couple of years on building infrastructure, and ensuring computational and data resources, the space is now focusing on delivering user-facing apps. AI agents are completely dominating right now, and it’s fascinating to observe how they’re developing, growing, and maturing.
Last week was marked by the 2025 edition of ETHDenver, one of the largest and most prestigious crypto events of the year. As usual, it gathered the hottest Crypto AI projects, so I figured, a summary of the most interesting talks is a good way for us to re-connect and re-enter the space.
Thank you for joining me again! Let’s dive in.
Building Autonomous Web3 Infrastructure: Secure Data & Decentralized Computing
This discussion was kicked off with the question: “What’s wrong with centralized AI” — a perfect entry point to the field of Decentralized AI and the raison d’être of this newsletter. The panel featured Valory‘s David Minarsch, Sydney Lai from Gaia 🌱, and Lit Protocol‘s David Sneider who provided their points of view on why we need to decentralize the control over AI.
Control is a major topic when it comes to AI, especially when we take into consideration that the field of artificial intelligence is currently dominated by an extremely limited set of companies. Who controls the models, the data they’re trained on, and the bias they inevitably produce? Answer these questions honestly, and you’ll get why decentralization is painfully needed.
Listen to the full conversation to find out what are the trust assumptions in decentralized AI and what assurances it provides in terms of ownership, data authenticity, and data monetization. It also touches on AI agents’ use in crypto, one of the most trendy topics right now.
The AI Agent Economy
As I mentioned, AI agents occupy an enormous part of crypto space’s mindshare at the moment, so the following fireside chat is worth checking out. It features Lit Protocol‘s David Sneider again, but this time in a conversation with CoinFund‘s CEO and founder Jake Brukhman. They break down what the AI agent economy, or agentic commerce, is, and what distinguishes agents from simple chatbots.
In their core, agents are software programs that can perform a certain task autonomously. Of course, they act on their owner’s behalf, but they have the agency to decide on the most efficient way to reach a pre-determined goal.
The first wave of AI agents that gained popularity over the last few months specialized mostly in summarizing and delivering information, mainly on X and Telegram Messenger. Wave 2 is gradually coming now with agents capable to automate other tasks, such as customer support, sales outreach, yield optimization, loss prevention, and authentication.
Most of the tech infrastructure powering agents has already been built, although outstanding problems still remain. Keeping context and memory is an open challenge in the science of artificial intelligence, as well as enabling agents to switch from one AI model to another.
Lastly, Brukhman boldly predicted that the open-source AI models born out of the Web3 space will be competitive with the centralized ones produced by OpenAI and Anthropic - something that many people strongly doubt. What do you think?
PublicAI Making AI More Accessible
Jordan Gray from PublicAI dedicated his ETHDenver talk to one pressing issue in AI: the lack of high-quality training data.
LLMs are great for satisfying the regular user’s needs, because they’re very generalized. When it comes to specific business applications, though, regular large language models struggle, and the reason hides in the lack of specified data. This is where Web3 comes in — where centralized data providers fail to deliver.
PublicAI works with 700,000 independent data builders, which helps them to achieve greater diversity in the data they can deliver, across demographics, nationality, and expertise. However, with such a high number of contributors, the data quality also varies.
To solve this problem, PublicAI employs slashing: a DeFi primitive used to guarantee the earnestness of blockchain validators. In this context, data builders are required to lock in a certain amount of tokens as an insurance against low-quality service. As long as they provide valid and high-quality data, their stake remains safe, and they earn more rewards.
Is this a killer app for Web3? Jordan thinks it is. To get truly aligned models, you need aligned incentives, and what other technology, besides blockchain, can really ensure that?
Why Every AI Agent Deserves a Wallet with Coinbase
Going back to the topic of AI agents, this conversation offers a nice overview of the recent growth of on-chain agents and the dawn of the so-called agentic commerce. Given the enormous potential, it’s no wonder that Coinbase is eager to be the one equipping all AI agents with a crypto wallet.
Explore this video to to see a live demo of Coinbase Developer Platform and learn how dApps’ interface will be progressing to become agent-friendly.
The Open Future of Robotics
It turns out that a real-life humanoid is already using a Coinbase wallet to buy and sell data and skills from other machines. Sounds surreal, right?
Meet Iris and her builder, OpenMind‘s Jan Liphardt. OpenMind’s objective is to create open-source software stack and operating system for robots — something like the Android for robotics. They’re also building an orchestration and communication layer, so robots can coordinate and interact with each other.
Watch to meet Iris - a fully autonomous humanoid, making history as the first robot with its own identity.
The Coming Agent to Agent Economy
Carra Wu from a16z crypto, Eskender Abebe from Eliza Labs, and Shi Khai WEI came together for a panel discussion on the challenges and opportunities of AI agents.
The panelists distinguished between agents for entertainment, including streaming and influencing, agents for code generation, and other types, all with different levels of autonomy. They also offered their predictions for how the space would look like in 10 to12 years. Taking into account the speed and dynamics of development, it’s almost impossible to forecast the future, but we can be fairly sure that agents will be widely used in digital twins, robotics, and even regulatory watchdogs.
Supercharging AI Possibilities: The Next Frontier in Crypto
Sreeram Kannan, founder of one of the most important crypto projects of the last few years, EigenLayer, made an attempt at answering the question: ‘Should we really pay attention to the Decentralized AI thesis and why?’ He described smart contract as the minimum viable agents, since they’re pre-programmed, unstoppable, and automating tasks on our behalf. But he also went one step further.
Can AI accelerate crypto subdomains like DeFi, DeSci, governance, and security? Not only that it can, but it’s already transforming the entire Web3 landscape. EigenLayer is building the Verifiable Commitment Cloud, a much broader project, having implications well beyond simple restaking. Find out below how it relates to AI.
Why 2025 is The Year Of The Agents
The last ETHDenver talk I selected to share with you is the one by NEAR Protocol‘s Illia Polosukhin. Before going into crypto, Illia was an AI researcher, and is one of the co-creators of the Transformer architecture, powering all large language models we’re using today.
Being among the fathers of modern AI, Polosukhin is a one of the strongest and most relevant voices in support of Decentralized AI. Listen to his lecture to find out what’s coming in the world of crypto agents and why he thinks we’re on the verge of a $1T market opportunity.
It’s worth noting that NEAR considers agents as the final component in reaching chain abstraction, i.e., removing blockchain complexities and making the Web3 ecosystem as user-friendly as the Web2 one.
Polosukhin also used the occasion to announce the Open Agents Alliance, featuring companies like Coinbase, Eliza Labs, Aethir, Overclock Labs, creators of Akash Network, Phala, Hyperbolic, Exabits.ai, and many others.
Bankless Live from ETHDenver with Delphi Digital
A couple of months ago, Bankless started developing the AI Rollup, a separate rubric, entirely dedicated to the burgeoning Crypto AI space. Its latest edition brought Delphi Digital‘s Tom Shaughnessy Jr, straight from ETHDenver, talking about their theses and arguments to be among the first to invest in this new subsegment of crypto over 2 years ago.
The obvious use cases they started supporting first were DePINs providing computational resources and distributed AI training projects. They eventually moved up the stack into middleware and inference, until eventually they reached the app layer with protocols like MyShell.ai.
Over the last quarter, the explosion of AI agents have exposed a bunch of new crypto users to Decentralized AI for the first time ever. Many of them don’t even know that this is just another wave of applications born in the intersection of blockchain and AI.
I recommend you listen to this conversation because it covers many important topics:
Why do we need utility agents that can perform valuable tasks on our behalf, and not just X spamming bots;
How will the tug of war between open-source and closed-source AI develop;
How can we solve outstanding problems like open-source data generation, and incentivization and monetization of open-source AI models;
What will the future of agents’ orchestration, curation, and interface look like;
How can we make sure that future apps will use a combination of models so we can guarantee a plurality of opinions and points of view;
How can we lower the steep barrier to entry and stimulate experimentation and R&D when AI is inherently highly resource intensive and expensive.
Thank you for reading! I hope you found this article insightful.
If you’d like to keep up with the latest developments at the burgeoning intersection of blockchain and AI, I invite you to subscribe.
And if you find value in my writing, consider supporting the newsletter. I’d greatly appreciate it.
The Web3 + AI Book Club is live on Fable. Follow the link below and come join our literary community.
I'm looking forward to connecting with fellow Crypto x AI enthusiasts, so don't hesitate to share your thoughts here or reach out on LinkedIn, Instagram, or Firefly.social.
Disclaimer: The content of this newsletter is provided for informational and educational purposes only. Nothing contained herein should be construed as financial advice or as a recommendation to buy, sell, or hold any of the companies or assets mentioned.
Please don’t take my views at face value. Instead, do your own research (DYOR), think critically, and share your perspective so we can challenge ideas, learn from one another, and arrive at better conclusions through thoughtful discussion.