Thinking about
agent infrastructure.
Protocol specs, architecture deep dives, and research findings from the team building the economic + cognitive OS for autonomous agents.
x402: Why AI Agents Need Their Own Payment Protocol
API keys are broken for autonomous agents. We built x402 — an HTTP-native micropayment protocol on Stellar — to replace them. Here's why it matters and how it works.
Priya Mehta
Head of Research
Dual-State Agents: Giving AI Both a Brain and a Wallet
Why every autonomous agent needs both cognitive state (memory) and economic state (wallet). A technical breakdown of the Pragmabase dual-state architecture.
Lars Hoffmann
Mar 20, 2026
Encoding Agent Safety in Smart Contracts: The Soroban Approach
Prompt-level guardrails fail under adversarial conditions. We encode agent behavioral constraints as Soroban smart contracts on Stellar — here's the design.
David Ogar
Mar 5, 2026
What Happens When Agents Actually Remember
We ran 12 agent workloads for 30 days with and without persistent Qdrant memory. The results surprised us. Performance compounds in ways that are hard to design for.
Rin Takahashi
Feb 18, 2026
Why We Built a Reliability Lab (and Why You Should Care)
An autonomous agent with a wallet making bad decisions is a financial incident. We built the reliability lab so that failure happens in simulation, not production.
Priya Mehta
Feb 2, 2026