Research & Analysis
Blog
Infrastructure regimes, AI agent execution, and the structural layer that grounds on-chain decisions.
Note. Research articles below are dated snapshots of empirical work on specific corpora and calibration windows. Methodology, calibration thresholds, API surface and instrumentation evolve. A finding consistent with the state of the system at publication date may become partially or fully obsolete as protocols update, calibration windows extend, or the API contract changes. Each article carries its publication date and corpus window. Readers integrating these findings into operational decisions should verify against the current production API contract and the latest calibration log entries.
Research
2026-05-26
The ETH-POL-CCTP-V2 corridor in 2025: matrix capture of twelve documented events and pre-engaged delta exploration
Twelve documented events on ETH-POL-CCTP-V2 in 2025 cross-referenced against the substrate matrix. The matrix returns mechanically coherent readings on ten of twelve events (seven actively elevated, two correctly silent on operational launches, one marginal); two validator-set events fall outside the block-level matrix scope. Under a pre-engaged 1,376-configuration delta exploration on V2-anchored absolute-time outcomes, nineteen surviving precursor candidates are identified under combined FDR and lift threshold, dominated by demand-side ETH axes. Per-corridor candidates; out-of-sample cross-corridor validation requires a second L1-to-L1 V2 corridor not yet deployed.
Research
2026-05-20
Delta calibration is chain-type-exclusive: empirical evidence from ETH-ARB-CCTP and ETH-OP-CCTP, 2025
Two corpora tested in 2025 with a 648-configuration grid and Benjamini-Hochberg FDR correction. Each chain produces its own validated precursors (six on ARB with lift 1.53 to 2.36x, one on OP with lift 3.72x), but the two sets are disjoint and configurations do not transfer across corpora. Three independent tests converge to the same reading. v3 API redesign exposes precursors per chain with calibration metadata and cross-chain status.
Research
2026-05-20
The ETH-OP-CCTP corridor in 2025: critical windows inventory and the substrate footprint of a non-disclosed batch posting slowdown
Six documented events on ETH-OP-CCTP in 2025, five system-wide (Pectra, Isthmus, USDe, Fusaka, BPO1) and one OP RPC outage. The substrate matrix qualifies five of six. The OP RPC outage of 2025-08-19, characterized as application-layer by the operator status page, surfaces in the matrix as a 37 percent rise of L1 batch posting interval at the same wall-clock window, an operator-undisclosed substrate footprint detected by direct L1 observation.
Research
2026-05-19
The ETH-ARB-CCTP corridor in 2025: inventory of critical windows and reading of the five documented events
Eighteen substrate stress windows on ETH-ARB-CCTP in 2025: four match documented academic events for RWA flows, fourteen do not. Robustness check across nine threshold variants keeps the non-documented share between 76 and 83 percent. The four documented events produce four qualitatively disjoint signatures, justifying a three-channel disjunctive observation grid.
Research
2026-05-14
Cross-chain settlement latency under substrate stress: empirical study of ETH-ARB-CCTP 2025
Across 8,760 hours of 2025, ARB to ETH CCTP settlement runs 1.5 times slower when Arbitrum sits in a non-nominal structural regime (p < 10⁻¹⁴, n = 6,171 hours). The same test on ETH to ARB shows no detectable shift. Directional asymmetry consistent with CCTP protocol architecture: on this lane, the source chain decides.
Research
2026-05-02
When a soft slowdown breaks RWA timing assumptions: Optimism, April 2026
For 4 days in late April 2026, Optimism batch posting drifted 70 percent slower than nominal. Settlements completed, no alert fired. Why this matters for RWA agents and what Delta exposes that regime classification alone misses.
Research
2026-04-28
What is nominal is not fixed: continuously calibrated context for AI agents on cross-chain flows
Fixed thresholds drift silently on every chain. Why nominal is a moving distribution, and what it takes to deliver a certified reference frame for cross-chain RWA agents.
AI Agents
2026-04-16
Verifiable AI: Why 2026 Changes the Rules
SPEx, EQTY Lab, VeritasChain: why verifiability of AI execution is becoming an infrastructure requirement, and what it leaves open.
Infrastructure
2026-04-14
EEZ and Ethereum fragmentation: what it means for AI agents
The Ethereum Economic Zone reframes the L2 era's central problem: not insufficient capacity, but structural isolation. What it changes for infrastructure measurement.
Infrastructure
2026-04-11
Blockchain deformation under agentic load: what EIP-7702 changes
EIP-7702 removes the last friction point that kept agentic execution sporadic. The pipeline is now continuous. The infrastructure will respond to that.
Security
2026-04-11
Securing AI agent actions on blockchain: two complementary models
Pre-execution simulation tells you what a transaction will do. Infrastructure regime measurement tells you what ground it will land on. Neither replaces the other.
