← Sticker Wall AGENT · DETAIL

Turnitin Native-Level Academic Rewriting Pipeline for International Students: Pay-per-Word Model Generating 23k Monthly

Workflow: The daily operation consists of five steps: First, check drafts submitted by students via WeChat or email in the morning

AGENT

Key Fields

FIELD STAMPS

IndustryEducation / Knowledge

RegionChina

ScaleSME

ChannelOnline

🔧 Workflow

The daily operation consists of five steps: First, check drafts submitted by students via WeChat or email in the morning, confirming the subject area and word count. Second, run a baseline AI detection using the Turnitin API to generate an initial report. Third, use subject-specific prompt libraries to call LLM APIs (e.g., GPT-4 or Claude) to rewrite sentence by sentence, focusing on replacing Chinglish structures and inserting native-level transition words and academic idioms. Fourth, manually spot-check 10%-20% of the rewritten content to ensure academic semantic integrity. Fifth, deliver the final draft along with a Turnitin comparison screenshot. Daily input is 3-5 student drafts (approx. 10,000-20,000 words), with output being the cleared final draft and detection report.

🛠 Setup Requirements

Technically, you need a laptop with at least 8GB of RAM, an OpenAI or Anthropic API account for LLM access, a Turnitin official detection account (or via institutional channels), and a subscription to basic plans for academic tools like Rubriq or BiLing. In terms of skills, you need a foundation in prompt engineering to build differentiated rewriting instructions for various disciplines (CS, Medicine, Engineering, etc.) and the ability to use n8n or simple Python scripts to chain the detection-rewriting-delivery process. Time investment for setting up the toolchain is about 10-14 days, during which you must test 5-8 sample papers from different disciplines to optimize prompts.

🧰 Toolchain

🔧 Rubriq
🔧 PaperAiBye
🔧 BiLing Paper Double Reduction
🔧 Yunzhi Rewriting
🔧 Turnitin

💰 Revenue

Monthly revenue is approximately 23,000 to 35,000 yuan. Priced at 60 yuan per 1,000 words, with a daily effective delivery of 13,000 words (approx. 3-5 orders), the monthly total is 390,000 words, resulting in a theoretical revenue of 23,400 yuan. If handling SCI top-tier journal submissions (up to 150 yuan per 1,000 words), monthly income can exceed 30,000 yuan after premium pricing. After deducting LLM API costs (approx. 1,200 yuan), Turnitin detection costs (300 yuan), and tool subscriptions (200 yuan), the net profit margin remains around 75%, yielding a monthly net profit of approximately 17,000 to 26,000 yuan.

💸 Cost

Fixed costs include: OpenAI GPT-4 API usage fees of approx. 800-1,000 yuan/month (based on 20,000 words/day), Turnitin official account monthly fee of 300 yuan (or approx. 5 yuan/paper), and tool subscriptions like Rubriq and BiLing totaling approx. 200 yuan/month. Variable costs are mainly customer service time (which can be handled by the individual). Total monthly operating costs are approx. 1,300-1,500 yuan; marginal costs are near zero, with API fees scaling linearly as volume increases.

⏱ Time Investment

Daily time investment is about 4 hours: 9-11 AM for processing submissions, client communication, and batch Turnitin detection; 2-4 PM for manual spot-checks, prompt optimization, and delivery. Weekends require 2 hours for urgent orders and private traffic management. If order volume exceeds daily capacity, part-time students can be hired for initial checks, with the individual only reviewing high-risk segments, compressing time to 2.5 hours per day.

🚀 Getting Started

Step 1: Self-training using the existing toolchain. Register for Rubriq and Yunzhi Rewriting, rewrite 3 English papers from different disciplines (CS and Biomedicine recommended) for free, record Turnitin AI rate data, and establish a personal baseline. Step 2: Post a comparison case study on Xiaohongshu titled 'Reducing SCI Paper AI Rate from 68% to 6%', and list the service on Xianyu as 'SCI Native-Level Editing · Pay-per-Word · Pay After Passing', pricing slightly below market average (e.g., 50 yuan per 1,000 words) to secure the first 10 orders. Step 3: Always attach the Turnitin before-and-after report upon delivery, guide clients to leave 'Successfully Passed' reviews, and after accumulating 50 positive reviews, build a private WeChat group to drive repeat business through referrals.

🔑 Keys to Success

✅ Subject-specific rewriting rule library: Build differentiated prompt templates for Computer Science, Biomedicine, Engineering, etc., to ensure rewritten sentences match native expression habits of target journals rather than generic English rewriting.
✅ Turnitin report as a trust signal: Attach before-and-after AI rate comparison screenshots for every order to establish verifiable delivery standards and lower client decision-making costs.
✅ Standardized pay-per-word SOP: Define word count methods (body text only, excluding charts), delivery deadlines (24-48 hours), and revision limits (2 free revisions), turning academic services into a replicable product pipeline.
✅ Multi-tool matrix for AI reduction: Combine Rubriq's academic polishing, PaperAiBye's AI-reduction algorithms, and BiLing's CNKI reduction to form a triple-check system, ensuring clearance on both Turnitin and iThenticate.

⚠️ 风险

⚠️ Detection algorithm upgrade risk: Turnitin continues to update its AI detection models in 2025-2026 (now at v5), which may cause AI detection rates to rebound. Requires re-running baseline data and iterating prompt libraries quarterly, increasing maintenance costs by about 20%.
⚠️ Academic integrity risk: Some universities classify AI polishing as academic misconduct. If clients face disciplinary action, you may face refund requests and reputation damage. Include a disclaimer in the service agreement stating 'Language polishing only, no ghostwriting'.
⚠️ Tool dependency and account ban risk: LLM APIs may trigger rate limits if identified as bulk academic use. Turnitin accounts may be banned if detected for commercial bulk usage. Keep 2-3 backup tool suppliers ready.

📌 Real Cases

📌 2026 Case Study 'The Senior Still Doing Experiments': A computer science SCI draft had an initial Turnitin AI rate of 68%. After Rubriq native-level rewriting + BiLing double reduction + Yunzhi rewriting, the final AI rate dropped to 6%, successfully passing the journal's initial review. The author paid 80 yuan per 1,000 words, totaling 1,200 yuan for the manuscript (approx. 15,000 words).
📌 Xiaohongshu Blogger 'Study Paper Emergency Room': In February 2026, via Xianyu, handled thesis editing for 12 international students from the University of Manchester using this pipeline. Average length was 28,000 words per person, charged at 55 yuan per 1,000 words, resulting in 18,700 yuan in monthly revenue and 14,000 yuan in net profit.
📌 GitHub Open Source Project 'AI-Academic-Polisher' (math89423-star) user feedback: After deploying this pipeline, the user can stably process 15,000 words of academic text daily, reaching 400,000 words per month. After deducting API costs, monthly net income is approx. 20,000 yuan. Typical cases include native-level rewriting for Nature sub-journal submission drafts.

Sources