Source-of-truth filing data for financial AI. Ready to integrate via API or MCP.

We turn complex SEC filings into structured, source-linked data built for financial AI. Your model handles generation. Lucy provides the source-grounded financial data behind it.

Trusted by Industry Leaders

AXS InvestmentsCanary CapitalRoboStrategyKraneSharesYieldMaxGraniteSharesVistaSharesAXS InvestmentsCanary CapitalRoboStrategyKraneSharesYieldMaxGraniteSharesVistaShares
Problems We Solve

Financial RAG fails when the source data is not trustworthy.

01

Hallucinations in Filing-Based RAG

Most hallucinations start before generation — when retrieved context is incomplete, mislabeled, or detached from its filing source.

02

Low Accuracy on Financial Calculations and Metrics

Numeric questions need exact figures, periods, and entity context. Without source-of-truth data, answers turn fragile and hard to verify.

03

Months Wasted on Preprocessing

Teams burn months on table extraction, XBRL grounding, and item segmentation before they can even start testing RAG quality.

Solutions We Provide

Lucy turns unstructured SEC filings into structured, retrieval-ready financial data.

Source-Linked Data

Every chunk traces back to the official filing source — retrieval that reduces hallucination at its root.

Prebuilt RAG Pipeline

Filing ingestion, parsing, chunking, and retrieval — fully assembled, ready to plug into your workflow.

One-Click Integration

Connect via MCP or API and retrieve source-backed filing chunks directly inside your AI stack. Lucy returns the evidence — your model composes the answer.

MCPREST APISource-linked
Benchmark Performance

Optimized for financial retrieval accuracy.

90%
Numeric answer accuracyRAGAS · FinQABench
15 min
Filing ingestion latencyNear-real-time ingestion.
Source-linked
Chunk traceabilityEvery chunk traces to source.

LucyRAG (Claude Fable 5) vs frontier models with Web Search

RAGAS answer accuracy

1.00.80.60.40.2
.99
.95
.85
.79
.79
gpt-5.6-sol+ Web Search
LucyRAGfable-5
gemini-3.6+ Web Search
fable-5+ Web Search
opus-5+ Web Search

Read the benchmark methodology

Proprietary Technology

Compress 200+ pages into 6 objects.

Text Blocks
Images
Financial Tables
XBRL Tags
References
Metadata

You need structured data, not a smarter model.

Table Extraction

✓Every row, column, and merged cell is preserved exactly as filed — no reconstruction guesswork.

XBRL Grounding

✓Each value is tagged with its official XBRL taxonomy label, eliminating ambiguity across filings and time periods.

Segmentation

✓LUCY segments by Item (1, 1A, 7, 8...) and sub-section, preserving the regulatory structure of each filing.

Chunking

✓Chunks respect document boundaries — tables, paragraphs, and exhibits are never split mid-element.
Coverage

What Lucy Covers.

37Filing types
2.9M+Processed
10 yearsHistorical Coverage
Live27 forms

Last updated: 2026-10-06 02:13 UTC

201620182020202220242026
Topics

Structured company intelligence from every filing.

  • Story01
  • Brief02
  • Products services brands03
  • Physical properties04
  • Income flow05
  • Acquisitions06
  • Customer concentration07
  • Insider trades08
  • Beneficial ownership09
  • Officers and directors10
Ready to get started

Build on Structured Financial Ground Truth

LUCY transforms raw financial documents into precision-structured data — so your AI does less guessing and more knowing.

Request Access