What’s real in Crypto + AI—and what’s just noise? In this episode of THE PEOPLE'S AI podcast, presented by Vana, we sat down with two top investors to unpack the actual state of decentralized AI. These are the people who see all the decks, hear all the pitches, and are funding what they believe is the future of the space. So what’s legit, what’s frothy, and where’s it all headed?

First, we talk with Alex Odagiu, Investment Director at YZI Labs. Alex gives a clear-eyed view of the current landscape, sharing insights on the wave of projects flooding the space, the use cases that excite him (think data markets, DeFi agents, and composable infrastructure), and the challenges of signal vs. noise. He also breaks down how he uses AI tools in his own VC workflow—practical, actionable insights for anyone trying to level up their research game.

Then we shift to a wide-ranging, deeply thoughtful conversation with Daniel Barabander, GC and Investment Partner at Variant. We get into the big picture of how crypto and AI fit together, especially around agent composability, economic ownership, and verification layers. He walks us through why AI agents might need to spend cryptocurrency, where that thesis holds up, and where it doesn’t. We also explore what’s still broken in the space—and what’s needed for real breakthroughs.

Topics include:

• [00:04:00] Why 75% of pitches Alex sees are Crypto + AI
• [00:10:00] Why blockchain incentives are well-suited for labeling scarce data
• [00:14:00] What’s broken in open-source model training—and how Web3 could fix it
• [00:18:00] Agents in DeFi: useful or meme coin mania?
• [00:21:00] How VCs vet AI x Crypto teams—and why founder ‘pivot mindset’ matters
• [00:25:00] Practical ways VCs use LLMs for diligence and synthesis
• [00:32:00] Daniel’s 3 pillars: Aggregation, Verification, and Self-Custody
• [00:36:00] Do AI agents really need to spend cryptocurrency?
• [00:44:00] Why economic ownership is crucial to decentralized AI
• [00:48:00] Composability as a path to superintelligence
• [00:55:00] What’s still broken: the “fuzzy verification” problem
• [01:00:00] The underrated promise of modular data layers

We close with a fun look at the modular vs. monolithic debate—Daniel makes the case for why the open, decentralized, and composable internet still has a shot.

If you’re at Consensus and still in Toronto, swing by the AI Summit and say hello.

Daniel Barabander:

https://x.com/dbarabander

Alex Odagiu:

https://x.com/odagius

And a big thanks to our partner, Vana, whose mission is to enable user-owned AI through user-owned data. Vana recently announced a collaboration with Flower Labs to build the world’s first user-owned foundation model.

About Vana's collaboration with Flower:

https://www.vana.org/posts/vana-flower-labs-partnership

About Vana:

https://linktr.ee/vanahq

Vana on Twitter/X:

https://x.com/vana

Vana ecosystem: vana.com

Subscribe to The People's AI on YouTube:

https://www.youtube.com/channel/UCnLiYlJulQIcmvCjnVRYotw

Jeff Wilser on Twitter/X:

https://x.com/jeffwilser

In this episode of The People’s AI, presented by Vana, we explore one of the most promising frontiers in artificial intelligence: federated learning.

We speak with Nic Lane, co-founder and Chief Scientific Officer of Flower Labs and a professor at the University of Cambridge, to unpack why decentralized AI may not only be more ethical—but more effective—than centralized systems.

We dive into:
•Why centralized AI may soon hit a wall due to limited data access
•How federated learning enables richer, more private, and more diverse datasets
•The mechanics behind training large language models across global networks
•Flower’s partnership with Vana to build Collective One, a new model powered by user-owned data
•Real-world use cases: from healthcare to weather forecasting to robotics
•Why decentralization isn’t just a principle—it might be a technical necessity

Whether you’re a developer, investor, or just curious about where AI is headed, this episode offers a lucid, technical-yet-accessible look at the future of open source AI, data DAOs, and user-owned AI models.

Subscribe on Apple, Spotify, or YouTube—and if you’re into crypto and AI, this is the conversation you don’t want to miss.

For more: https://flower.ai/

Presented by Vana, the open protocol for user-owned AI through user-owned data. Vana's vision is for user-owned AI through user owned-data. Its mission is to be the world's first open protocol for data sovereignty.

https://linktr.ee/vanahq

Vana on Twitter/X:

https://x.com/vana

Vana ecosystem: vana.com

Vana at MIT Decentralized AI Summit, Builders Workshop

Subscribe to The People's AI on YouTube:

https://www.youtube.com/channel/UCnLiYlJulQIcmvCjnVRYotw

Jeff Wilser on Twitter/X:

https://x.com/jeffwilser

In this episode of The People’s AI, presented by Vana, we dive into one of the most compelling intersections in tech today: AI agents and crypto infrastructure. Our guest is Bryan Pellegrino, co-founder and CEO of LayerZero Labs. His journey is anything but ordinary—former top-ranked professional poker player, model builder for MLB teams alongside Billy Beane, and now a key architect of decentralized AI infrastructure.

We explore the deep mechanics of decentralized agents: How will they live on-chain? Will they need wallets? How will reputation systems work for non-human actors? And most critically—what does it mean for blockchains when agents, not humans, drive the majority of transactions?

Bryan unpacks why micro-payments are broken in legacy systems and how crypto rails solve that. He discusses why LayerZero should be thought of as the TCP/IP of the decentralized internet—abstracting cross-chain operations for AI agents and making assets truly fungible across blockchains.

We also tackle big-picture topics like:

• [06:10] The problem with overhyping LLMs and the limits of scaling compute
• [14:45] What LayerZero actually does and why it matters
• [17:40] AI agents as blockchain-native actors—and why payments come first
• [27:30] The challenge of AI hallucination vs. blockchain immutability
• [31:20] Wallets, security, and reputation for AI agents
• [37:10] Use cases for decentralized AI people are sleeping on
• [44:45] How to build infrastructure that can outlast rapid AI evolution
• [46:00] Bryan’s grounded, long-view predictions for AI, AGI, and agent proliferation

Bryan brings rare insight—and skepticism—to the table, showing where hype diverges from real engineering, and where massive opportunity still lies. If you care about decentralized AI, agentic systems, or the future of crypto infrastructure, this one’s not to be missed.

For more: Layer Zero Labs

Presented by Vana, the open protocol for user-owned AI through user-owned data. Vana's vision is for user-owned AI through user owned-data. Its mission is to be the world's first open protocol for data sovereignty.

https://linktr.ee/vanahq

Vana on Twitter/X:

https://x.com/vana

Vana ecosystem: vana.com

Vana at MIT Decentralized AI Summit, Builders Workshop

Subscribe to The People's AI on YouTube:

https://www.youtube.com/channel/UCnLiYlJulQIcmvCjnVRYotw

Jeff Wilser on Twitter/X:

https://x.com/jeffwilser