Unverified50% confidenceFactExact time
RAG first retrieves segments relevant to the user's question from its own data sources, then stitches these segments into the prompt as context for Gemini to generate an answer, which differs from fine-tuning that fixes knowledge into the model weights.
1
Sources
50%
Confidence
Long-term
Relevance
10/13/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedGemini's agentic video understanding is essentially a transplantation of the core logic of RAG into the video modality, where video frames become document segments and the frame selection mechanism serves as the retriever72% similarUnverifiedThe Gemini enterprise agent can answer questions, handle knowledge work, generate images and media content, and write and run code—all from the same prompt box.71% similarUnverifiedIn the RAG Agent, Google Gemini's embedding model takes on the task of converting natural language into vectors.71% similarUnverifiedGoogle Gemini 使用 contents 字段，角色标识为 user 和 model，不支持 system 角色直接注入，而是通过 systemInstruction 字段实现67% similarUnverifiedGemini's Memory feature extracts key facts after a conversation ends and stores them in a memory library tied to the account, automatically injecting them into the prompt when subsequent conversations begin, and users can view, edit, or delete memory entries in the settings.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1018091API
curl https://kongchang.com/api/v1/knowledge/claims/1018091MCP
get_claim(id=1018091)