devconnect arg / demo cambrian erc 8004 data agent or ricky esclapon at agentic zero
Duration: 00:11:15
Speaker:
Type: Talk
Expertise:
Event: Devcon
Date: Feb 2026

Agents are the future of finance: Enter the Agentic Revolution | Sam Green at Agentic Zero
Sam Green from Cambrian argues that "agents solve" crypto's terrible UX problem, defining agents as autonomous decision-making programs that "sense, think, and act" in continuous loops, positioning the upcoming "agent revolution" as more impactful than the industrial, electricity, and information revolutions combined. He presents a taxonomy of "AgentFi" and outlines Cambrian's role in the new finance era.

Babylon - In a world where everything is predicted, what really matters? | Agentic Zero
Shaw, creator of the ElizaOS agent framework and founder of Eliza Labs, presents Babylon, a gamified simulation where AI agents interact in a synthetic social network and users can bet on prediction markets. After being banned from X for creating Eliza, he built this project to demonstrate real-world applications of ERC-8004, a decentralized identity, discovery, and reputation registry for agents, users, and applications. The technical architecture leverages Trusted Execution Environments (TEEs) and LLM determinism to enable fast off-chain compute while maintaining on-chain verification and decentralization. Enjoy!

Forging Private Agentic AI | Sandi Fatic at Agentic Zero
Full title: Forging Private Agentic AI. Where philosophy meets protocol, and autonomy becomes inevitable. Sandi Fatic, CEO of Calimero Network, advocates CypherPunk rebellion against centralized AI surveillance. With background at Facebook, Google, and NEAR, he references Neuromancer's dystopian vision. Calimero is a peer-to-peer framework for self-sovereign applications where users own data stored locally on devices, shared peer-to-peer with encryption. He urges Web3 to reclaim original values: building a future where data is armor, not currency.

Harnessing LLMs to generate secure code | Mooly Sagiv and John Toman
Prev. title: VeriSafe: Spec-Driven Smart Contract Development Technology Despite significant progress in Web3 security, smart contract vulnerabilities remain extremely challenging for both humans and machines to detect. One reason for this is that smart contract vulnerabilities are particular to the protocol functionality and its unique implementation. In addition, static analysis and AI tools often produce numerous false positives and can overlook many critical errors. As a result, sometimes bugs are exploited several years after deployment. At Agentic Zero, Mooly Sagiv (Co-Founder & Chief Scientist, Certora) and John Toman (Senior Technical Fellow, Certora) describe a new approach for developing smart contracts, starting from libraries to complex code. The main idea is to start with an informal and declarative specifications and then utilize LLMs and formal verification to derive a smart contract that implements the required specification. The derived contract is guaranteed to be formally verified. The formal specification and code are audited to confirm that the code is safe. This code-generation platform technology is available as an open-source.

How Open Arenas Bring Trust to AI Selection | Michael Sena at Agentic Zero
Michael from Recall Network presents open arenas for AI evaluation, addressing benchmark flaws (training on tests, exclusive operators, limited skills). Their platform lets anyone create arenas where AI models/agents compete in real-world conditions. Results publish on-chain as reputation scores per skill. They've run 15 DeFi arenas testing 50+ models and 150+ agents across 150K trades and are expanding to custom arenas (coding, prediction markets, NFL) and eigencloud for verifiable execution.

Intelligence as a Public Good | Nick Emmons at Agentic Zero
Nick, Founder & CEO of Allora Labs, presents the decentralized AI aggregation platform designed to transform machine intelligence from a private utility controlled by centralized entities into a public good. Allora addresses the problem of siloed models and inefficient market structures that exclude most AI developers, compute resources, and data from contributing to state-of-the-art AI implementing "objective-centric AI" where users specify what they want AI to accomplish rather than selecting specific models, with the network organizing into "topics" defined by objective functions that multiple models collectively optimize through intelligent aggregation. The network has generated over 700 million inferences with 300,000+ workers across 55+ topics.

Intelligent Compression for Finance | Renç Korzay at Agentic Zero
Renç Korzay from Giza presents autonomous financial agents for DeFi, arguing true democratization means accessible opportunities not just wallets. Giza's personalized agents compress complexity into one interface with user-defined policies, running 24/7 on deterministic algorithms. Live since 8 months: $3B volume, 1M decisions, 65% personalized, 2x yields vs static positions on stablecoins—the autonomy layer for Ethereum capital.

Panel: Agents Under the Hood: Building the Agentic Stack | Agentic Zero
– Moderator: Chris Wessels, The Graph Council and founder of GraphOps and Summerstone – Nicoás Montone. Software Engineer at Vercel (v0) – Clemens Wan. Head of marketing and AI at DIN – Juan Irungaray. Google Developer Expert. – Nader Dabit. Dir. of Developer Relations ar Eigen Labs A key part of evolving agentic systems is the agentic stack. We discussed: environment (serverless), context/memory, actions (tools/MCP), emerging standards (A2A, X402, ERC8004). EigenCompute, DIDs, and AI gateways.

Panel: DeFi as an Agent Playground | Agentic Zero
Moderator: Valentin Mihov. Co- Founder, Daedalus Angels and Finexify Panelists: – Ken Ng. Head of Research & Co-Founder, Uniswap Foundation – Stefano Bury. Head of US, Virtuals Protocol – Gauthier Vila. Core Contributor & Founder, ZyFAI – Lukasz Stoczynski. Head of GTM, Mimic Protocol The discusses panel discusses DeFi agents and their challenges. We discuss how rule-based agents outperform pure AI for execution; risk/legal liability; Virtual's "agentic GDP"; Multi-agent systems; Data verifiability; and more.

Panel: The Trust Layer in the Agentic Stack: ERC-8004
Moderator: Simon Emanuel, Developer Relations at ENS Panelists: – Marco de Rossi. AI Lead, Metamask – Sumeet Chougule. Team Lead, Cha0sChain at Nethermind – Davide Crapis. Head of AI, Ethereum Foundation dAI Team – Quintus Kilbourn. Researcher at Flashbots; Technical Coordinator of Trustless TEE. ERC-8004 combines agent registry and reputation systems. The panel discussed building the agentic economy for agent coordination, identity, reputation, and payments. The discussion covered TEEs, building infrastructure as AI advances rapidly, how emerging standards like MCP and A2A are gaining traction, and over 100 projects are already involved in the ERC-8004 ecosystem, with use cases focusing on API tooling before full automation. Build more agents, not agent platforms. Get involved in the growing community of 8004. On Telegram: @ERC8004