GrepSeek DCI

I want to evaluate and, if it fits, integrate a technique from recent AI research into my project.

## Technique: GrepSeek DCI
GrepSeek: Direct Corpus Interaction — Hybrid Search for AI Agents

Drop-in applicable: DCI bypasses vector embeddings for exact-match retrieval (error codes, version constraints, file paths) while semantic retrieval handles broad recall — GrepSeek fuses both via RRF achieving nDCG@10=0.847 (+19% vs pure semantic RAG). Direct POC: implement a hybrid search demo with BM25 semantic layer + regex DCI layer + RRF fusion over a corpus containing both natural-language docs and exact identifiers.

What the reference demo shows: Gradio demo with three search modes (semantic / DCI / hybrid) over a corpus of AI engineering documents — showing exactly where semantic search fails on exact strings and DCI recovers them.

## Reference implementation (Vector Ridge Labs, POC-012)
- Live demo: https://osoman2-gradio-pocs.hf.space/?poc=POC-012
- Source code: https://huggingface.co/spaces/osoman2/Gradio_pocs/tree/main/grepseek_dci
Original source: AlphaSignal newsletter deep-dive · 2026-05-31

## My project
<Describe your stack, the problem you want to solve and where this technique would sit. Example: "Python FastAPI RAG service over 50k PDFs, Postgres + pgvector, OpenAI embeddings, p95 latency budget 800 ms.">

## What I need from you
1. Read the reference source code and explain in at most 5 bullets how the technique works and what it would change compared to my current approach.
2. Decide whether it fits my case. If it does not, say why and stop there.
3. Propose the smallest viable integration: files or modules to touch, new dependencies, configuration.
4. Implement it behind a flag or as a separate function so it can be compared side by side with what I have today.
5. Add an evaluation script that measures retrieval quality (recall@k / nDCG on a labelled sample of my queries) and latency, before and after, on my own data.
6. List the risks (latency, cost, failure modes) and what to monitor once it is in production.

Constraints: keep my existing stack and conventions, avoid heavy dependencies unless the gain is clear, and reference the file in the demo source whenever you reuse its logic.