Answers are grounded in this paper and cite their sources.
Computational study demonstrates automated seismic hazard calculations across seventy-three cities, highlighting accessible end-to-end hazard modeling via language model agents.
Probabilistic seismic hazard and risk analyses are the backbone of building codes, insurance pricing, and disaster management. Yet their pipelines remain accessible primarily to experts. We present, to our knowledge, the first agentic interface to the end-to-end probabilistic seismic hazard and risk chain via an open-source server, addressable through the Model Context Protocol (MCP). MCP wraps the OpenQuake engine and the 2020 European Seismic Hazard and Risk Models using twenty-four typed endpoints. An agent is a large language model that chooses which tools to call and in what order, reads their results, and revises its plan. LLM is confined to the role of an orchestrator, while the OpenQuake engine and custom codes compute hazard, damage, and loss. Each response carries source-model, ground-motion model (GMM), and data provenance. Benchmarked against ESHM20 at seventy-three cities, replicated 475-year spectral accelerations match the published values within a median of 5%, and within 28% at 97% of cities. The interface accepts user-supplied empirical or machine-learning GMMs on any tectonic region type, and adds features existing services omit: conditional spectra, deterministic scenarios, surface hazard, per-building loss, retrofit comparison, and record selection. The design is meant to transfer to other regional models and hazards.
No takes yet. Share an insight, caveat, or question.
Vemula et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: