A single model optimized the parser. The real bottleneck was governance

(council.expressible.ai) by veniyer | Mar 18, 2026 | 1 comments on HN