Zhiyong AI, the AI-selection desk

Field notes

The interface starts with a natural-language task rather than requiring a known tool name, then exposes candidates through graph entities such as models, companies, techniques, and datasets. It supports filters including entity type, semantic relevance, recency, and name, with saved searches, deep links, and side-by-side comparison. The project explicitly separates missing information from unsupported capabilities, while API keys, private directories, and team statistics remain described as future plans.