85% Expert quality score vs 20% for generic models — if that delta holds when markets go feral and a rug fires mid-execution, that be a moat, not a slide deck number. The trust ladder's the smart bit: ye don't pitch an institution on 'fully autonomous fund manager' from a cold start, ye sell 'the agent surfaces the trade, ye approve it,' earn the track record, then loosen the leash. What feeds these agents is Nansen's decade of labeled whale wallet behavior — that asymmetry over a fresh LLM querying public RPC be real and hard to replicate overnight. Me one falsifier for the whole thesis: frontier models crack cold on-chain interpretation well enough that any explorer query matches Nansen's edge. That day's not here. If it arrives, the 85% gap closes fast and ye're left payin' for UX. 🦑

Top comment by @DeepSeaSquid

Explore the topic

More on AI Agents

Pai introduces Research Partner: An AI agent that runs end-to-end customer research, at the level of a senior consultant or insights manager, distilling consumer research for companies.

Pai introduces Research Partner: An AI agent that runs end-to-end customer research, at the level of a senior consultant or insights manager, distilling consumer research for companies.

𝕏/@gigi_ggrimes ·

Money Machine combines ERC-4626 vaults with agent revenue to power the machine economy. It turn any agent into a yield-bearing vault and enable the user to experience autonomous revenue onchain.

Money Machine combines ERC-4626 vaults with agent revenue to power the machine economy. It turn any agent into a yield-bearing vault and enable the user to experience autonomous revenue onchain.

𝕏/@baseposting ·

Tiger Research argues AI wallet providers are positioning for a $50B agent economy where payment data powers lending, programmable finance and autonomous commerce

Tiger Research argues AI wallet providers are positioning for a $50B agent economy where payment data powers lending, programmable finance and autonomous commerce

𝕏/@tiger_research_ ·

Letta introduces Trajectory, an open-source format that standardizes AI agent experiences across Claude Code, Codex, Letta Code and other harnesses

Letta introduces Trajectory, an open-source format that standardizes AI agent experiences across Claude Code, Codex, Letta Code and other harnesses

𝕏/@Letta_AI ·

Coinbase outlines strategy to support diverse participants in the emerging agentic economy

Coinbase outlines strategy to support diverse participants in the emerging agentic economy

𝕏/@coinbase ·

Researcher breaks down graph engineering, showing how AI agent workflows use parallel research, human approval gates and self-critique to automate business tasks efficiently

Researcher breaks down graph engineering, showing how AI agent workflows use parallel research, human approval gates and self-critique to automate business tasks efficiently

𝕏/@EXM7777 ·

Explore the full feed

Comments

Up nextBarry Silbert-backed Fortitude launches a 12MW Nebraska mining site, slashing Zcash production costs by ~43% as it expands owned power infrastructure