Unverified50% confidenceFactExact time
The core goal of Hillock v1.0 is to let the system clearly identify and refuse unanswerable queries, rather than returning plausible-looking fragments.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/28/2026
First Seen
Valid until: 1/26/2027
Sources
Related Entities
Related Claims
UnverifiedWhen a question cannot be answered by verified facts, Hillock's gating mechanism returns a hard-coded rejection response within 1 millisecond without invoking Ollama, thereby preventing large model hallucinations at the architectural level and saving GPU compute overhead.73% similarUnverifiedHillock is not positioned to replace vector databases, but to introduce deterministic judgment at the retrieval layer.69% similarUnverifiedHillock adds a layer of judgment beyond similarity: whether the retrieval result truly constitutes effective support for the query.67% similarUnverifiedRAG's anti-hallucination measure is to use system prompts to constrain answers to come solely from retrieved documents, requiring the AI to respond that there is no relevant information when the information is not present in the context.64% similarUnverifiedIn DSupport, when the RAG results are empty after filtering, the tool returns nothing_relevant_found, and the knowledge expert is required to answer NO_ANSWER rather than fabricate an answer.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1106476API
curl https://kongchang.com/api/v1/knowledge/claims/1106476MCP
get_claim(id=1106476)