Structured Data Extraction from D&B Credit Reports using GPT-4o - n8n Workflow

Credit Risk Analysts who need to standardize data from dozens of D&B reports.
Automation Engineers looking for advanced n8n templates for PDF parsing.
Finance professionals seeking to integrate Dun & Bradstreet data directly into internal systems (CRMs, databases).
Developers building complex data pipelines using the flexibility of an n8n node sequence.

Manually reading and transcribing complex financial documents like D&B Company Reports is time-consuming and prone to human error. These reports are often delivered as PDFs, making automated data ingestion difficult. This sophisticated n8n workflow solves this problem by chaining several specialized n8n node operations: authentication, PDF retrieval, text extraction, and advanced AI-driven parsing.

This specific n8n template uses GPT-4o, configured with a strict extraction system prompt and a defined JSON schema, to reliably pull fields like PAYDEX scores, composite credit appraisals, and viability scores directly into a clean, structured JSON object. This robust approach ensures high accuracy and consistency, transforming unstructured documents into actionable data ready for use in any downstream n8n node integration.

This comprehensive n8n workflow operates in a precise sequence to handle both API integration and complex document processing:

1. Authentication and Token Retrieval: The flow begins with the Get Token n8n node (HTTP Request) which authenticates against the D&B API using Basic Auth credentials. It retrieves a required accesstoken to be used for subsequent API calls.


2. Report Fetching: The D&B Report n8n node (HTTP Request) uses the retrieved token to call the D&B API, requesting a specific company's report in PDF format (using the reportFormat=PDF parameter).


3. File Preparation: The raw API response is piped into the Convert to PDF File n8n node, which ensures the binary data is correctly identified as a PDF file type within the n8n workflow.


4. Text Extraction: The Extract Binary n8n node utilizes its PDF operation capability to read the binary file and extract all readable text content from the financial report.


5. AI Analysis (The Core n8n Node): The extracted text is passed to the powerful Analyze PDF n8n node (LangChain Agent). This node receives highly specific instructions via a system prompt to act as a "precision extractor," focusing solely on key financial scores and metrics.


6. Structured Output Definition: The output is routed through the Structured Output n8n node, which enforces a mandatory JSON schema (including fields like paydex3mo, dnbratingoverall, and creditlimitconservative). This critical step, supported by dedicated OpenAI Chat Model n8n node instances for both analysis and auto-fixing, guarantees the final output adheres exactly to the required structure, making this n8n workflow reliable for data ingress.

To use this n8n template, follow these steps:

1. Import: Copy the provided JSON and import it into your n8n instance via the 'Workflows' screen using the 'New' button and 'Import from JSON' option.


2. OpenAI Credentials: Set up credentials for the OpenAI Chat Model n8n node instances (OpenAI Chat Model6 and OpenAI Chat Model7). You will need an API key for GPT-4o access.


3. D&B Authentication: Configure the Get Token n8n node:

1. D&B Report URL: In the D&B Report n8n node, verify the URL and headers are configured to dynamically use the access token received from the Get Token n8n node and the desired DUNS number. Note that this specific n8n workflow uses a hardcoded DUNS number in the URL example, which you should replace with an expression referencing an upstream n8n trigger input or custom variable.


2. Execution: Once configured, execute the n8n workflow to test the data extraction process.

Get Token (HTTP Request n8n node): Function: Secures an access token from the D&B API using Basic Authentication. Key Configuration: Uses POST method, URL https://plus.dnb.com/v3/token, and body parameter granttype=clientcredentials.
D&B Report (HTTP Request n8n node): Function: Fetches the raw D&B Company Report PDF using the dynamically retrieved Bearer token. Key Configuration: GET request specifying reportFormat=PDF.
Convert to PDF File (n8n node): Function: Transforms the raw binary output from the API request into a recognized file object within the n8n workflow environment.
Extract Binary (n8n node): Function: Specifically extracts the textual content from the internal PDF file object, preparing it for AI analysis. Key Configuration: Operation set to 'pdf'.
Analyze PDF (LangChain Agent n8n node): Function: The primary processing n8n node that applies GPT-4o intelligence to read and interpret the complex financial text. Key Configuration: System message provides extremely detailed instructions for flat JSON extraction and scoring criteria.
Structured Output (LangChain Output Parser n8n node): Function: Enforces the schema for the output data, ensuring the AI response strictly conforms to the required fields (e.g., paydex24mo, viabilityscore). This makes the output data highly usable by subsequent n8n node operations.

• OpenAI Chat Model6/7 (n8n node): Function: Provides the large language model capabilities (GPT-4o) necessary for the advanced analytical and parsing tasks managed by the Agent and Output Parser n8n nodes.