---
title: "Discussion on \"Why Your RAG Answers Still Hallucinate in .NET: Root Cause and Fix\" | Hashnode"
description: "Discussion on \"Why Your RAG Answers Still Hallucinate in .NET: Root Cause and Fix\". RAG still hallucinating in .NET? The retrieval threshold, grounded prompt, and citation fixes that force honest answers and a real I don't know."
url: "https://hashnode.com/posts/rag-grounded-answers-citations-dotnet/6a747749e2d0ef0754fa2fe6"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [Why Your RAG Answers Still Hallucinate in .NET: Root Cause and Fix](https://codingdroplets.com/rag-grounded-answers-citations-dotnet?utm_source=hashnode&utm_medium=feed)

RAG still hallucinating in .NET? The retrieval threshold, grounded prompt, and citation fixes that force honest answers and a real I don't know.

Aug 6 · 11 min read · codingdroplets.com  
Read the article

-   [#net](https://hashnode.com/tag/net)
-   [#csharp](https://hashnode.com/tag/csharp)
-   [#ai](https://hashnode.com/tag/ai)
-   [#llm](https://hashnode.com/tag/llm)
-   [#rag](https://hashnode.com/tag/rag)
-   [#aspnet-core](https://hashnode.com/tag/aspnet-core)
-   [#backend](https://hashnode.com/tag/backend)
-   [#api](https://hashnode.com/tag/api)
-   [#dotnet](https://hashnode.com/tag/dotnet)

## Written by

[Coding Droplets](https://hashnode.com/@codingdroplets)

@codingdroplets

  
[View profile](https://hashnode.com/@codingdroplets)

Build, Scale & Master .NET with Clean, Practical Tutorials

## Responses 4

Markdown supported · ⌘↵ to post

[Puneet Khandelwal](https://hashnode.com/@lifestyle4u)

[Aug 8](https://hashnode.com/posts/rag-grounded-answers-citations-dotnet/6a747749e2d0ef0754fa2fe6/comment/6a770925575732cd699895fa)

Always interesting to see how different approaches to data tracking can shape our understanding of ourselves

[Coding Droplets](https://hashnode.com/@codingdroplets)

[Aug 17](https://hashnode.com/posts/rag-grounded-answers-citations-dotnet/6a747749e2d0ef0754fa2fe6/comment/6a770925575732cd699895fa#reply-6a832e17e4871b5bde169965)

Thanks!

[Kartik N V J K](https://hashnode.com/@kartik-nvj)

[Aug 6](https://hashnode.com/posts/rag-grounded-answers-citations-dotnet/6a747749e2d0ef0754fa2fe6/comment/6a74bd4a2088f57d32831faf)

The point that a model handed context will use it whether or not it's relevant is the root cause people keep prompt-engineering around instead of fixing upstream. Two independent refusal layers is the part I'd emphasize, because a single relevance check tends to fail the same way the retriever did. Citations that carry their own scores are underrated too, since a low-confidence citation the user can click beats a fluent answer with nothing behind it.

[Coding Droplets](https://hashnode.com/@codingdroplets)

[Aug 17](https://hashnode.com/posts/rag-grounded-answers-citations-dotnet/6a747749e2d0ef0754fa2fe6/comment/6a74bd4a2088f57d32831faf#reply-6a832df1b9c829555a56b4f2)

Hi Kartik, Thanks! You picked the part I'd underline too.

The correlated-failure point is exactly why one check isn't enough. If the relevance filter uses the same embeddings as the retriever, it agrees with the retriever's mistake by construction. Second layer has to be judging something different.

On citations, the scores earned their keep in support more than in the product. When someone reports a bad answer, the scores tell you instantly whether it was retrieval or generation.
