Customer work · Finance
SEC filings in, compensation data out.
Palm Hill built Ebullience a document-intelligence platform that reads SEC filings and extracts executive compensation into structured records, using language models, Azure Document Intelligence, and retrieval over the filings.
Ebullience’s site; opens in a new tab. Ebullience is a Palm Hill customer; the platform is theirs.
What was needed
Executive compensation sits inside long SEC filings, in tables and prose that change from company to company. Ebullience needed it as structured data: records that can be searched, compared, and loaded into other systems, rather than pages that someone has to read and retype.
What Palm Hill built
- Read the filing. Azure Document Intelligence turns each SEC filing into text and tables.
- Find the right passages. Retrieval narrows the filing to the parts that discuss compensation.
- Extract and structure. Language models pull executive compensation out of those passages into structured records.
- Store and show. Results are kept in MongoDB and SQL Server and presented in a Blazor web application.
Stack
- Language models
- Azure Document Intelligence
- Retrieval (RAG)
- MongoDB
- SQL Server
- Blazor
- Python
- Docker
- .NET Aspire
Services run in Docker and are composed with .NET Aspire; the extraction work is written in Python.
Have filings or contracts to structure?
The same document work is available as an engagement. We test it on samples of your own documents before we build.