Introducing Clairic, a new System One class of extraction models

Engineering teams building data ingestion pipelines frequently squander latency and budget by forcing general-purpose generative models into predictable information-extraction roles. Cloudflare has addressed this mismatch with Clairic, a specialized System One extraction model designed to extract, compute, estimate, and forecast structured data directly from unstructured inputs. Rather than leaning on heavyweight reasoning chains for operational parsing, the model focuses on rapid, high-throughput interpretation. For technical operators, this design moves complex data ingestion out of slow generative loops and into efficient, edge-ready pipelines. This architecture offers immediate operational leverage for systems burdened by unstructured text. A freight brokerage startup in Chicago, for instance, can pipe fragmented dispatch emails and non-standard bills of lading through an extraction endpoint to pull route stops, compute dynamic fuel surcharges, and forecast delivery windows in a single step, bypassing brittle chains of custom parsers. In Austin, an outpatient medical administration