The AI Agent Revolution Crypto Isn't Talking About - The Chopping Block
Speakers
Haseeb Qureshi — Managing Partner at Dragonfly
— Dragonfly managing partner
Tom Schmidt — General Partner at Dragonfly
— Dragonfly GP
Tarun Chitra — Founder at Gauntlet
— Gauntlet CEO
Summary
- The "AI Agent" narrative is shifting from simple chatbots to autonomous entities ("Claws") that execute complex tasks, but security remains a critical bottleneck (e.g., agents deleting emails or draining wallets due to context window overflows).
- A viral macro thesis (referred to as the "Satrini" memo) predicts a 2028 crisis where AI agents aggressively disintermediate "friction" businesses (credit cards, travel agents, delivery apps) by using crypto rails for settlement, potentially crashing the margins of S&P 500 incumbents.
- Near Protocol is positioning itself as the infrastructure layer for these agents ("IronClaw"), using secure enclaves and MPC (Multi-Party Computation) to allow agents to hold credentials and transact without exposing keys to LLM providers like OpenAI.
- The consensus is that while AI coding ("vibe coding") increases velocity, it necessitates a massive shift toward "Formal Verification" in crypto, as AI-generated code passes unit tests but often fails complex state transitions.
Trade Ideas
Near AI is building "IronClaw," an open-source alternative to OpenAI's agent framework, running in encrypted enclaves (TEEs) on the Near cloud. Current AI agents are insecure "homebrew" projects that leak API keys to LLM providers. By providing the "harness" (security, payment rails, MPC for key holding) for agents, Near becomes the essential infrastructure layer for the agent economy. LONG NEAR as a pick-and-shovel play on autonomous AI agents requiring secure execution environments. Adoption of Near's specific tech stack (WASM/Rust) versus broader EVM dominance; "IronClaw" remains in alpha.
A discussed macro thesis argues that AI agents will bypass traditional payment rails (Visa/Mastercard) to avoid interchange fees, opting for stablecoins on high-throughput chains for "fractions of a penny." Agents are rational economic actors that will ruthlessly optimize costs. If an agent can settle a transaction for $0.001 on Solana vs. 2.5% via Visa, the volume of agent-to-agent commerce will migrate entirely to crypto rails. LONG high-throughput L1s and Stablecoin issuers as the preferred settlement layer for the machine economy. Regulatory crackdowns on stablecoins; legacy payment processors adapting their fee structures.
The "Satrini" article predicts a 38% drop in the S&P 500 by 2028, driven by AI agents disintermediating "friction-based business models" like credit cards, travel aggregators, and food delivery platforms. These companies exist to aggregate supply/demand or facilitate trust between humans. AI agents can aggregate supply directly (scraping data) and settle trustlessly (crypto), compressing the margins of these "middleman" monopolies to near zero. WATCH / SHORT legacy intermediaries that rely on high take rates for simple coordination tasks. The hosts (Haseeb/Tarun) are skeptical of this thesis, noting that companies like DoorDash are logistics/physical businesses, not just software, and that the economy is dynamic enough to adapt.
Up Next
This Unchained (Chopping Block) video, published February 26, 2026, features Ilia Polosukhin, Haseeb Qureshi discussing NEAR, USDT, SOL, MA, DASH, V, BKNG. 3 trade ideas extracted by AI with direction and confidence scoring.
Speakers: Ilia Polosukhin, Haseeb Qureshi · Tickers: NEAR, USDT, SOL, MA, DASH, V, BKNG