LLM Accuracy Evaluation Against Reference Answers in Le

This workflow automatically evaluates the accuracy of LLM-generated responses in the legal domain by comparing them to reference answers and storing Pass/Fail verdicts with reasoning in Google Sheets. Ideal for testing AI system precision, monitoring model regression, and benchmarking performance across different LLMs.

## Who it´s for
- AI developers needing to test LLM accuracy
- Legal startups using AI for document processing
- QA teams automating text generation evaluation
- Researchers comparing performance of different LLMs

## What the automation does
- Triggered manually or via HTTP webhook
- Fetches test cases from Google Sheets: inputs, LLM outputs, and reference answers
- Sends each LLM output to a language model via OpenRouter for evaluation
- Returns a Pass/Fail decision with explanation based on comparison to reference
- Appends results — verdict and reasoning — back to the same spreadsheet

## What´s included
- Ready-to-use n8n workflow with LangChain agent
- Trigger logic (manual and HTTP webhook)
- Integrations with Google Sheets, OpenRouter, and external APIs via HTTP
- Basic textual guide for setup and adaptation

## Requirements for setup
- n8n instance (self-hosted or cloud)
- Google Sheets account with read/write permissions
- OpenRouter API key
- Webhook hosting (e.g., Railway) if using HTTP triggers

## Benefits and outcomes
- Automates repetitive LLM quality checks
- Ensures consistent evaluation against a fixed reference
- Transparent decisions with explanatory reasoning
- Centralized tracking of results in a shared spreadsheet
- Enables A/B testing of different models via OpenRouter
- Supports QA and development processes for legal AI tools

## Important: template only
Important: you are purchasing a ready-made automation workflow template only. Rollout into your infrastructure, connecting specific accounts and services, 1:1 setup help, custom adjustments for non-standard stacks and any consulting support are provided as a separate paid service at an individual rate. To discuss custom work or 1:1 help, contact via Telegram: @gleb923.