Evaluates LLMs on legal test cases from Google Sheets using GPT-4 as judge, scoring pass/fail with reasons and logging results back to Sheets.
This n8n workflow automates the benchmarking of large language models (LLMs) on a dataset of legal information extraction tasks. It pulls test cases from a Google Sheet, processes each by retrieving source documents via Google Drive, converting them to text, and submitting the LLM's input/output to a GPT-4.1 'judge' via OpenRouter for evaluation (Pass/Fail + reasoning). Results are appended to an output sheet in real-time, with a 0.5s pause to respect API rate limits.
Key benefits include effor
Platform
n8n
Category
IT & Development
Price
$16.99
Creator
Jordi Faber
LLM
evaluation
benchmarking
Google Sheets
OpenAI
Google Drive
AI testing
analytics
legal AI
automation
How to import this workflow into n8n
1Purchase or download the workflow to get the n8n workflow JSON file.
2In your n8n instance, open Workflows and choose "Import from File" (or paste the JSON with Ctrl+V on the canvas).
3Open each node marked with a credential warning and connect your own accounts and API keys.
4Run the workflow once manually to verify the data flow, then toggle it to Active.