Grass founder says AI inference needs regional node coverage
- —Radonjic: inference cannot wait, unlike collecting data for model training
- —AI agents typically fetch 5–50 pages per web lookup, more for deeper reasoning
- —Contents API returns page content via Grass nodes with a selectable country
- —Search API is the next product; launch moved up after infrastructure optimization
Why it matters: Grass is extending its home-bandwidth model from training data to live AI agent queries, raising demand for its GRASS token.
Source: TokenPost