Business Analyst Skill (11/10 Enhanced)

To extract exact financial figures like Total Revenue from a 10-K filing without hallucinations, this Skill uses high-precision regex for structured financial extraction. This method eliminates number hallucinations by directly isolating GAAP metrics from the document.

You can analyze both qualitative strategic risks and quantitative financial data in a single 10-K query. The Skill combines adaptive retrieval for complex strategic analysis like competitive moats with structured extraction for precise revenue metrics.

You need Python and LangGraph to run 10-K analysis with local models. The implementation requires Ollama models specifically Deepseek-R1 and Nomic-Embed, alongside potentially external APIs for web search fallback.

Adaptive retrieval is a mechanism that intelligently routes simple 10-K queries for faster responses while performing deep RAG for complex analysis. It ensures efficient processing by matching retrieval depth to query complexity.

To handle missing or outdated information in SEC filings, the business analyst skill employs self-correcting analysis. It uses document grading and a web search fallback to retrieve supplementary data, ensuring robust and accurate outputs.

The best way to verify extracted Net Income and EPS data against source 10-K documents is using the built-in hallucination checking mechanism. This validates generated responses against the original filing text to ensure accuracy.