Beancount.io LogoBeancount.io

Bean Labs Research Log

Open experiments and findings from Bean Labs — the Finance AI Agent research initiative by Beancount.io. Browse by tag.

OpenHands: Open Platform for AI Software Agents and What It Means for Finance Automation
·mike

OpenHands: Open Platform for AI Software Agents and What It Means for Finance Automation

OpenHands is an MIT-licensed, Docker-sandboxed agent platform where CodeAct achieves 26% on SWE-Bench Lite — a sobering benchmark that establishes what AI agents can reliably do today, and why the first productive finance deployments should be tightly scoped rather than autonomous.

ai
open-source
automation
llm
+4
Fin-RATE: How LLMs Fail at Cross-Period and Cross-Entity Financial Analysis
·mike

Fin-RATE: How LLMs Fail at Cross-Period and Cross-Entity Financial Analysis

Fin-RATE benchmarks 17 LLMs on 7,500 expert-curated QA pairs from 2,472 SEC filings, revealing an 18.60% accuracy collapse under longitudinal tracking and a 54-point drop for finance-specialized Fin-R1 on cross-entity tasks — with the retrieval pipeline, not the backbone model, as the binding bottleneck.

llm
ai
machine-learning
analytics
+3
FinDER: Real Analyst Queries Expose a 74% Recall Gap in Financial RAG
·mike

FinDER: Real Analyst Queries Expose a 74% Recall Gap in Financial RAG

FinDER benchmarks RAG on 5,703 real hedge fund analyst queries against S&P 500 10-K filings; E5-Mistral achieves only 25.95% context recall, and abbreviation-heavy queries cost 8.2 precision points — evidence that query normalization, not better embeddings, is the first fix for finance AI pipelines.

ai
llm
machine-learning
finance
+3
Lost in the Middle: Position Bias in LLMs and Its Impact on Finance AI
·mike

Lost in the Middle: Position Bias in LLMs and Its Impact on Finance AI

The TACL 2024 paper by Liu et al. shows LLMs perform up to 20 points worse on information buried in the middle of long contexts — a U-shaped degradation affecting every tested model including Claude-1.3-100K — with concrete implications for how RAG pipelines should order retrieved passages in finance and accounting applications.

llm
ai
machine-learning
data-science
+3
AD-LLM Benchmark: GPT-4o Hits 0.93+ AUROC Zero-Shot for Text Anomaly Detection
·mike

AD-LLM Benchmark: GPT-4o Hits 0.93+ AUROC Zero-Shot for Text Anomaly Detection

AD-LLM benchmarks GPT-4o and Llama 3.1 8B across three anomaly detection roles — zero-shot detector, data augmenter, and model selector — on five NLP datasets; GPT-4o reaches AUROC 0.93–0.99 zero-shot, but LLM-based model selection remains unreliable, with direct implications for financial audit AI.

llm
ai
machine-learning
data-science
+3
CausalTAD: Causal Column Ordering for LLM Tabular Anomaly Detection
·mike

CausalTAD: Causal Column Ordering for LLM Tabular Anomaly Detection

CausalTAD improves LLM-based tabular anomaly detection by reordering table columns to respect causal dependencies before serialization, lifting average AUC-ROC from 0.803 to 0.834 over AnoLLM on mixed-type benchmarks — with direct implications for detecting anomalies in structured ledger data.

llm
ai
machine-learning
fraud-detection
+3
AnoLLM: Fine-Tuning LLMs for Tabular Anomaly Detection in Financial Data
·mike

AnoLLM: Fine-Tuning LLMs for Tabular Anomaly Detection in Financial Data

AnoLLM (ICLR 2025) reformulates tabular anomaly detection as LLM density estimation — fine-tuning on normal rows and scoring by negative log-likelihood. It outperforms classical methods on mixed-type fraud datasets but offers no edge on purely numerical data, with real implications for detecting anomalies in Beancount ledger entries.

ai
llm
machine-learning
fraud-detection
+3
LLMs Score 2.3% on Beancount DSL Generation: The LLMFinLiteracy Benchmark
·mike

LLMs Score 2.3% on Beancount DSL Generation: The LLMFinLiteracy Benchmark

The LLMFinLiteracy benchmark finds that five open-weight ~7B models generate fully correct Beancount transactions only 2.3% of the time, with failures concentrated in accounting reasoning—not syntax—pointing to compiler-in-the-loop feedback as the critical missing ingredient for reliable write-back agents.

llm
beancount
plain-text-accounting
ai
+4
TableMaster: Adaptive Reasoning for Table Understanding with LLMs
·mike

TableMaster: Adaptive Reasoning for Table Understanding with LLMs

TableMaster is a prompting-only pipeline that reaches 78.13% on WikiTQ with GPT-4o-mini—13 points above Chain-of-Table—by combining table-of-focus extraction, semantic verbalization, and adaptive switching between text and symbolic reasoning. Here is what the architecture means for AI agents over financial ledgers like Beancount.

ai
llm
machine-learning
beancount
+4
Zero-Shot Anomaly Detection with LLMs: How GPT-4 Performs on Tabular Data
·mike

Zero-Shot Anomaly Detection with LLMs: How GPT-4 Performs on Tabular Data

GPT-4 achieves 74.1 mean AUROC on the ODDS benchmark without fine-tuning — nearly matching the classical ECOD baseline at 75.5 — but fails on multi-dimensional anomalies and high-variance datasets; a critical review of zero-shot LLM anomaly detection and its implications for automated Beancount ledger auditing.

ai
llm
fraud-detection
machine-learning
+3
DocFinQA: Long-Context Financial Reasoning on Full SEC Filings
·mike

DocFinQA: Long-Context Financial Reasoning on Full SEC Filings

DocFinQA replaces FinQA's curated 700-word passages with full 123,000-word SEC filings, exposing a 175× context increase that nearly halves GPT-4 accuracy on long documents. Retrieval pipelines fail to surface the right chunk 45% of the time at HR@3 — and long-context models are not a substitute.

ai
llm
machine-learning
finance
+3
TheAgentCompany: Benchmarking LLM Agents on Real-World Enterprise Tasks
·mike

TheAgentCompany: Benchmarking LLM Agents on Real-World Enterprise Tasks

TheAgentCompany tests 175 real workplace tasks across a simulated intranet with GitLab, OwnCloud, and RocketChat. The best model (Gemini-2.5-Pro) completes only 30% of tasks at $4 each, revealing that autonomous agents remain far from viable for accounting and finance workflows.

ai
llm
automation
machine-learning
+3
Showing 13–24 of 89 posts