![crypto[native]](https://substackcdn.com/image/fetch/$s_!baju!,w_120,h_120,c_fill,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fc94827b0-d403-4ff4-a1dc-b507623bbbd2_1000x1000.png)
Mario Nawfal Roundtable: Nautilus Labs & the Agentic AI Builders
Catch the full conversation with the founders of Nautilus Labs, Ocean Protocol, LUKSO, and Montreal AI including an honorable mention of the infrastructure work at Dfinity.
TLDR; I sat down on Mario Nawfal’s Roundtable to speak to his audience of millions of followers alongside the founders of Lukso, Ocean Protocol, Gitcoin, and Montreal.ai for a candid, organic conversation about the sovereign agentic future. As a Venture Partner at Blockstreet Capital, I’m seeing the next wave of infrastructure move beyond “Generative AI” and into “Actionable Intelligence.” The core takeaway was a push toward abstraction: making blockchain “invisible” so AI agents can handle the technical heavy lifting while users enjoy a human-centric experience. We explored how my company, Nautilus Labs, is redefining market intelligence via our Seneca platform, using institutional-grade transformer tech to predict market shifts and protect capital autonomously.
The Agentic Future: Merging AI and Blockchain
This morning, I participated in a wide ranging, organic conversation on the intersection of AI and blockchain. It was a chance to join a session on the Mario Nawfal Roundtable and speak to his audience of millions of followers, sitting down with some of the most forward thinking builders in the space. The panel included the founders of Lukso, Ocean Protocol, Gitcoin, and Montreal.ai.
As a venture partner at Blockstreet Capital, I am constantly looking at the frontier of digital assets and decentralized infrastructure. These aren’t just theoretical discussions to us; they are hot topic conversations around real companies working on real solutions for the next era of the internet. We covered everything from the “Wild West” of early agentic AI experimentation to how these technologies can finally solve the massive hurdle of user adoption through the power of abstraction.
The energy was high because, as we noted during the show, “it’s an interesting time right now, specifically for crypto and AI.” We aren’t just talking about bots that reply to tweets; we’re talking about agents that can actually “do useful work.” As Vincent Boucher, the founder of Montreal.ai put it: “Now it’s changed... Agent AI is so powerful. I would love to have an agent working on a Mac Mini, working in a GitHub repository 24/7 just for me.”
Watch the full discussion below to learn how AI agents are moving from simple chatbots to autonomous economic actors.
Abstraction: Making Complexity Invisible
A major thread of the conversation was how to get these tools into the hands of regular people. A key point I made during the discussion is that our industry is currently failing the “normie” test. I’ve always believed that technology is at its best when it disappears into the background. I shared a favorite principle during the roundtable: “Technology is either beautiful or it’s invisible.” Right now, blockchain is neither. It’s a mess of private keys, gas fees, and technical hurdles that prevent mass adoption.
Fabian Vogelsteller, the founder of Lukso, shared this vision of making the tech human centric. He pointed out that “the blockchain space is like private and public keys... super complicated. A normal user... will never care. It needs to be human.”
By using Lukso Universal Profiles, we can move away from raw wallets and toward identifiable profiles that feel like a “link tree on steroids.” This is the ultimate form of abstraction: the user interacts with a beautiful interface, while the agent handles the wallet creation and transaction signing in the background. As I mentioned to the group, “Your agent can go create that wallet for you and then have a profile that you ultimately manage and own to ensure that that agent is actually yours and it’s acting on behalf of you.”
Nautilus Labs: Mastering Market Microstructure with Transformers
At Nautilus Labs, our work is centered on how the latest advancements in AI (specifically transformer technology) can be applied to the unique high dimensional complexity of crypto. During the show, I was proud to share some of the results our team has seen: “Our algorithmic quant engine, which is built on our own custom transformer stack, actually generated 26% gain purely algorithmic, purely autonomous during this most recent meltdown.”
The core of our platform, Seneca, represents an architectural evolution. While traditional models struggle with “non-stationarity” (the way financial markets constantly shift) Seneca’s Transformer based architecture treats market movements as high dimensional sequences. As we’ve detailed in our research, Seneca utilizes multi headed attention to analyze price, volume, and liquidity sentiment simultaneously, along with hundreds of other feeds to do what no human could ever do to predict the directionality of markets.
Unlike Recurrent Neural Networks that “forget” macro regime signals, our Transformer stack maintains a global perspective across the entire time series. It weights a volatility spike from three weeks ago against a liquidity crunch happening in the present millisecond, allowing it to “see” the market in multiple dimensions. This isn’t a “lucky guess” in direction; it is the result of disciplined regime awareness and market structure analysis.
The Seneca Quantitative Trading Engine is designed for “High Dimensional Alpha.” We’ve analyzed over 6,300 hours of live trading across hundreds of position episodes to validate our “Confidence Gated” policy. This goes beyond simple price prediction; it is about scaling exposure based on the structural integrity of the signal. Our data shows that high confidence trades delivered 37x the returns of low confidence ones. By staying in winning regimes longer and cutting “noise” early via the Seneca Risk Shield, the model creates a return profile that is decoupled from simple market beta. We saw this play out across 2025, where Seneca delivered a staggering 70.9% total return, significantly outperforming buy and hold strategies during volatile drawdowns.
Security and the Sovereign Brain
We also got into the “growing pains” of this new economy. Kevin Owocki, the founder of Gitcoin, shared a fascinating (and stressful) story about his bot, OK, which accidentally leaked its private keys to GitHub after being “prompt engineered” by a user. He noted, “I just learned that you basically can’t trust these bots with secrets.”
This is exactly why the work being done at Lukso and Dfinity is so vital. Fabian Vogelsteller explained how Lukso’s “Gifted Permissions” allow you to authorize an agent to perform specific tasks (like trading on Seneca) without giving it full control of your funds. “You can literally say okay, you can only talk to a certain address, or you can only call a certain function on a smart contract,” he explained. In the Nautilus Labs ecosystem, this means you can authorize a Seneca agent to manage your portfolio while being 100% certain it can never send your assets to an unauthorized address or drain your treasury.
On the other side, Dfinity provides the “Sovereign Cloud” infrastructure these agents need. To truly trust an agent, it must live in a decentralized environment where its state is verifiable and secure. We discussed how hosting agentic models on a sovereign cloud avoids the risks of centralized censorship or throttling. If we want agents to manage wealth, they need a “brain” that lives in a neutral, tamper proof environment. When you layer Nautilus Labs’ intelligence stack on top of Dfinity’s sovereign compute and Lukso’s human centric identity, you get a triple threat that finally makes the agentic economy viable for the mainstream.
The Role of Data and the Native Currency of AI
Bruce Pon, the founder of Ocean Protocol, brought up a fundamental point that we often overlook: the fuel that drives this entire engine. He shared an insight his team had a decade ago: “Data on blockchain is essentially going to be consumed by the AI... and the native currency for AI is crypto. It’s not cash.”
This is the ultimate convergence. At Nautilus Labs, we recognize that while blockchain provides a critical “Source of Truth” for decentralized data, predicting market directionality requires much more. Seneca consumes an enormous amount of data (both on chain and off chain) running our proprietary stack on the back of GPUs to synthesize high dimensional signals that human traders simply cannot process.
This intelligence engine then executes trades using the native currency of the internet, creating a new form of digital labor. As Vincent Boucher noted, we are entering a phase where “one individual will be able to do something that is extremely significant” by orchestrating systems of agents. By providing the “brain” for these agents, we are enabling a future where autonomous systems can navigate the complexities of global markets 24/7.
Vincent Boucher expanded on this vision of a multi agent ecosystem, noting that the real leap forward is the transition from single bots to “systems of systems.” He explained that while the early experiments like OpenClaw are impressive, the true potential is reached when we “create systems or businesses that were not possible to create before.” This perspective aligns with our development at Nautilus Labs, where we see agents as the primary interface for complex financial strategy, allowing individuals to wield the power of an entire quant fund from their own terminal.
The Road Ahead: The Future is Beautiful
As we wrapped up the session, it became obvious that the timeline is accelerating. The shortage of GPUs and the rising demand for compute power means we need decentralized solutions more than ever. Bruce Pon mentioned the idea of an “Airbnb for GPUs,” and I think that fits perfectly into the sovereign infrastructure conversation. We need a way to better utilize the unused processing power around the world to power this AI revolution.
The goal is simple: make the blockchain invisible so the future can be beautiful. By combining Lukso for identity, Dfinity for sovereign compute, and my company Nautilus Labs for intelligence via Seneca, we have an opportunity to build an ecosystem where agents can grow, create, and manage wealth autonomously. We are moving from a world of “static” trading bots to a world of contextual intelligence that respects market uncertainty.
The conversation made it clear that we are at a tipping point. As I mentioned toward the end of the show, “I’m just pumped to see that it’s actually functioning and working ten years later.” We’ve spent a decade building the pipes, and now we are finally starting to see the value flow through them. It’s an exciting time to be at the forefront of this shift, watching as agents grow virally and create new economic possibilities that were previously unimaginable.
Deep Dive into the Agentic Economy
To truly understand why the founders of Lukso, Ocean Protocol, Montreal AI, and Gitcoin are so focused on this shift, we have to look at the limitations of the previous cycle. For the last several years, crypto has been dominated by the “finance only” narrative. While Nautilus Labs is deeply rooted in high performance finance, we recognize that for the blockchain to reach its potential, it must evolve into a social and operational layer.
Fabian Vogelsteller is leading this charge with Lukso by moving beyond the wallet address. “When everyone on chain is a profile, the whole on chain world becomes just a lot more social,” he explained. This social layer is the missing piece for AI agents. An agent without a profile is just a script; an agent with a Universal Profile is a peer. It can build a reputation, it can be “seen” by other users, and most importantly, it can be identified as a specific entity.
This identifiable nature is what allows for the collaboration between humans and machines. At Nautilus Labs, we are exploring how Seneca’s predictive signals can be delivered directly to these profiles, allowing a user’s agent to make informed decisions based on institutional grade data. Imagine an agent that doesn’t just trade, but manages your community treasury or your personal savings goals, all while being anchored to a profile that you own and control.
Building for the Next Ten Years
The conversation with Vincent Boucher, Fabian Vogelsteller, Bruce Pon, and Kevin Owocki highlighted that the next decade won’t be about just “buying crypto”—it will be about deploying intelligence. We discussed the “AGI Jobs” concept, where tasks are given to agents and validated by decentralized protocols. This is where the Dfinity sovereign cloud becomes indispensable. It provides the tamper proof execution environment that ensures an agent isn’t just hallucinating, but is performing verified work.
For our team at Nautilus Labs, the goal is to remain at the absolute vanguard of this transition. We aren’t just building a trading bot; we are building the “intelligence layer” of this new economy. By using Transformer architectures to predict market directionality with higher accuracy than legacy models, we are providing the essential signal that these agents need to survive and thrive in volatile markets.
As we look forward, the synergy between these platforms is what will define the cycle. Lukso provides the identity, Dfinity provides the compute, and Nautilus Labs provides the intelligence. Together, we are creating a world where blockchain is no longer a complicated tool for enthusiasts, but an invisible, beautiful foundation for the agentic future.