---
title: "Reading Is Not Using: Retrieval, Judgment, and the Design of AI Financial Research Workflows"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2608.24842v1"
permalink: "https://feed.astralor.com/items/itm_fada2bd17b75ca77ea9b4e65"
published: "2026-08-25T17:31:25.000Z"
score: 78
channel: "ai"
tags: ["10-K filings","AI financial analyst","Large Language Models","Workflow Design","Retrieval-Augmented"]
---

# Reading Is Not Using: Retrieval, Judgment, and the Design of AI Financial Research Workflows

## AI commentary

> This paper hits a blind spot in current AI finance products: we often score models by retrieval accuracy while forgetting whether they actually use that information in decisions. Teams building investment-research agents should pay special attention to this retrieval-integration gap—your system may pinpoint risk clauses yet completely ignore them in final judgments. I recommend auditing your workflow now for similar information eviction effects.

## At a glance

This paper identifies a retrieval-integration gap in LLM-based financial analysis: even when models accurately retrieve risk disclosures, those disclosures may fail to influence investment judgments. By varying unrelated context from 2,000 to 128,000 tokens while holding focal-firm information fixed, the authors find that the influence of a risk disclosure on investment judgments falls to the experimental noise floor even as direct retrieval remains accurate. The pattern replicates across model families, judgment tasks, and experiments removing real disclosures from actual 10-K filings. Causal memory interventions show that compressed summaries and source-text lookup jointly transmit disclosures into judgments, yet workflow architecture determines whether transmission succeeds: chunk-and-summarize pipelines evict relevant information, while a targeted, structured restatement adjacent to the decision restores its influence. The study concludes that AI analyst performance is jointly determined by model capability and workflow architecture, and retrieval-based evaluations can certify systems whose investment judgments ignore information they demonstrably retrieved.

## Key points

- Identifies a retrieval-integration gap: in long contexts, models can retrieve risk disclosures yet their judgments are unaffected.
- Robust across model families, judgment tasks, and experiments with real 10-K filings.
- Causal memory interventions show compressed summaries and source-text lookup jointly transmit disclosures, but workflow architecture determines integration success.
- Chunk-and-summarize pipelines evict relevant information, while targeted structured restatement near the decision restores influence.

## Full text

Large language models (LLMs) are increasingly deployed as AI analysts to process financial disclosures and support AI-assisted investment decisions. Yet such systems are usually evaluated by what they can retrieve, not whether retrieved information affects their judgments. We identify a retrieval-integration gap in long-context financial analysis. Holding focal-firm information fixed and varying only unrelated context from 2,000 to 128,000 tokens, we find that a risk disclosure's influence on investment judgments falls to the experimental noise floor even as direct retrieval remains accurate. The pattern replicates across model families and judgment tasks and in experiments removing real disclosures from actual 10-K filings. More capable models postpone but do not eliminate the gap. Causal memory interventions show that compressed summaries and source-text lookup jointly transmit disclosures into judgments. Workflow architecture determines whether this transmission succeeds: chunk-and-summarize pipelines evict relevant information, whereas a targeted, structured restatement adjacent to the decision restores its influence. AI analyst performance is therefore jointly determined by model capability and workflow architecture. Retrieval-based evaluations can certify systems whose investment judgments ignore information they demonstrably retrieved.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_fada2bd17b75ca77ea9b4e65