ragflow / index.html
Yingfeng Zhang
Rebrand page for Context as Infrastructure paper; remove unused template assets
e90220a
Raw History Blame Contribute Delete
5.89 kB
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<meta name="description"
content="Knowledge compilation for retrieval in agentic systems with harnesses.">
<meta name="keywords" content="RAG, agentic retrieval, knowledge compilation, context engineering, RAGFlow">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>Context as Infrastructure: Knowledge Compilation for Retrieval in Agentic Systems with Harnesses</title>
<link href="https://fonts.googleapis.com/css?family=Google+Sans|Noto+Sans|Castoro"
rel="stylesheet">
<link rel="stylesheet" href="./static/css/bulma.min.css">
<link rel="stylesheet" href="./static/css/fontawesome.all.min.css">
<link rel="stylesheet"
href="https://cdn.jsdelivr.net/gh/jpswalsh/academicons@1/css/academicons.min.css">
<link rel="stylesheet" href="./static/css/index.css">
<link rel="icon" href="./static/images/favicon.svg">
<script defer src="./static/js/fontawesome.all.min.js"></script>
</head>
<body>
<section class="hero">
<div class="hero-body">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column has-text-centered">
<h1 class="title is-2 publication-title">Context as Infrastructure: Knowledge
Compilation for Retrieval in Agentic Systems with Harnesses</h1>
<div class="is-size-5 publication-authors">
<span class="author-block"><sup>1</sup>InfiniFlow</span>
</div>
<div class="column has-text-centered">
<div class="publication-links">
<!-- arXiv Link. -->
<span class="link-block">
<a href="https://arxiv.org/abs/2610.41172" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="ai ai-arxiv"></i>
</span>
<span>arXiv</span>
</a>
</span>
<!-- Code Link. -->
<span class="link-block">
<a href="https://github.com/infiniflow/ragflow" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fab fa-github"></i>
</span>
<span>Code</span>
</a>
</span>
</div>
</div>
</div>
</div>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-desktop">
<!-- Abstract. -->
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title is-3">Abstract</h2>
<div class="content has-text-justified">
<p>
Retrieval-augmented generation is becoming agentic. A search harness writes queries,
reads results, reformulates, and decides when to stop. This largely solves accuracy: a
harness with only BM25/grep and MiniMax M3 answers 85% of BrowseComp-Plus and 91% on
multi-hop FRAMES in our runs, without compiled or vector indexes. It exposes a context
problem: an undisciplined loop repeats queries, rereads documents, and fills the window
with unused passages.
</p>
<p>
The missing layer is knowledge compilation: a configurable ingestion pipeline where LLMs
enrich documents with context no chunk contains alone&mdash;navigation hierarchies with
claim-evidence leaves, entity graphs, timelines, mind maps, ontologies, wikis, and memory.
</p>
<p>
Compilation is not a better retriever: under a competent harness, structural artifacts
gave no detectable accuracy gain on FRAMES, and a PageIndex-style outline degraded at
scale. It is a supply layer. A navigation tree whose leaves are claim-evidence
pairs&mdash;the mechanism we measure&mdash;cuts query tokens by 91% when evidence
suffices and adds 1.9% when it does not, a 38% average reduction; this implies a 2%
break-even sufficiency rate and a measured rate near 43%.
</p>
<p>
A decision gate cuts latency from 140 s to 30 s, though substitution's token saving is
bought with retrieval rounds and carries its own latency cost.
</p>
<p>
Other artifacts&mdash;ontology, graph, wiki, timeline, mind map, memory&mdash;pay off in
deployment (governance, integration, onboarding, temporal orientation, continuity) where
accuracy benchmarks cannot see them.
</p>
<p>
We describe RAGFlow's pipeline and harness (open source at
<a href="https://github.com/infiniflow/ragflow" target="_blank">github.com/infiniflow/ragflow</a>),
which turns the loop into five disciplined stations and keeps the context entering the
model the smallest sufficient set; together they form an agent-native database: a store
that hands the agent compiled, governed context instead of a ranked list of passages.
</p>
</div>
</div>
</div>
<!--/ Abstract. -->
</div>
</section>
<section class="section" id="BibTeX">
<div class="container is-max-desktop content">
<h2 class="title">BibTeX</h2>
<pre><code>@misc{ragflow2026context,
author = {{InfiniFlow}},
title = {Context as Infrastructure: Knowledge Compilation for Retrieval
in Agentic Systems with Harnesses},
year = {2026},
eprint = {2610.41172},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.41172}
}</code></pre>
</div>
</section>
</body>
</html>