Detect, Analyze and Humanize 🕵️‍♂️🔎 AI-generated text using Rephrasy

This workflow is designed to analyze and humanize AI-generated text automatically using Rephrasy AI. The workflow then sends this text to an AI detection service to evaluate how “AI-like” it is. The text is broken down into individual sentences, and each sentence is scored based on its likelihood of being AI-generated. Using a conditional logic step, the workflow identifies sentences with a high AI probability (score > 50). Only those sentences are sent to a humanization API, which rewrites them in a more natural, human-like style. Meanwhile, sentences that are already considered human are left unchanged. Finally, all processed sentences are recombined into a single output text. 👉 In short, the workflow: Detects AI-generated content Filters only the parts that need improvement Rewrites them to sound more human Reassembles the final optimized text --- Key Benefits 1. ✅ Selective Humanization Instead of rewriting the entire text, the workflow only modifies sentences that are likely AI-generated. This preserves natural content and reduces unnecessary changes. 2. ✅ Cost Optimization By processing only flagged sentences through the humanizer API, it minimizes API usage and costs. 3. ✅ Improved Text Quality The final output sounds more natural and less robotic, improving readability and engagement. 4. ✅ Sentence-Level Precision The workflow analyzes text at a granular level, allowing highly targeted improvements rather than bulk rewriting. 5. ✅ Automation & Scalability Once set up, the process is fully automated and can handle large volumes of text without manual intervention. 6. ✅ Multi-language Support The workflow supports multiple languages (even though the humanization step is set to English), making it adaptable to different use cases. 7. ✅ Maintain Structure & Meaning Since only specific sentences are modified, the original structure and intent of the text remain intact. --- How it works This workflow is designed to process a block of text, detect which sentences are likely AI-generated, and rewrite only those sentences to sound more human-like. 1. Trigger and Input: The workflow starts manually. A "Set text" node allows the user to define the text to be processed (stored in the text field). 2. AI Detection: The text is sent to the Rephrasy AI Detector API. The API returns an overall AI score and a breakdown (sentences) with individual scores for each sentence. 3. Data Parsing: A "Code" node transforms the API's sentence data into a structured array, tagging each sentence as "ai" (if the score > 50) or "human". 4. Splitting and Looping: The "Split Out" node separates the sentences into individual items. The "Loop Over Items" node then iterates over each sentence to decide if it needs processing. 5. Conditional Humanization: For each sentence, the "> 50?" node (IF condition) checks the AI score. If the score is greater than 50 (indicating AI), the workflow routes the sentence to the AI Humanizer node to be rewritten. If the score is 50 or less (human), the sentence is passed through unchanged. 6. Reassembly: Once the loop finishes, the "Aggregate Text" node collects all the sentences (both the rewritten AI sentences and the original human sentences) and reassembles them into a single, final text output. --- Set up steps To get this workflow running, you need to configure the connections to the Rephrasy API and define the input text. 1. Configure Credentials: The workflow uses two nodes: AI Detector and AI Humanizer. Both require an HTTP Bearer Auth credential. You must create a credential in n8n named "Rephrasy" (or update the node settings to match your credential's name) and populate it with your valid Rephrasy API key. 2. Set Input Text: Locate the Set text node. Modify the text assignment value. Currently, it contains a placeholder (XXX). Replace this with the text you wish to humanize, or configure it to reference data from a previous trigger (e.g., from a webhook or form). 3. Review Language Settings: In the AI Humanizer node, check the language parameter inside the JSON body. It is currently set to "English". Ensure this matches the language of the text you are processing. 4. Activate Workflow: Once the credentials are set and the input text is defined, toggle the workflow from Inactive to Active (or simply execute it manually using the "Execute Workflow" button if no trigger is needed). --- 👉 Subscribe to my new YouTube channel. Here I’ll share videos and Shorts with practical tutorials and FREE templates for n8n. []( --- Need help customizing? Contact me for consulting and support or add me on Linkedin.