← Sticker Wall AGENT · DETAIL

Netomi E-commerce Email and Chat Auto-Resolution Side Hustle: 8,000 to 20,000 Yuan per Month

Workflow: Every day, receive customer complaint text from emails or chats forwarded by sellers. The system first uses a large mode

AGENT

Key Fields

FIELD STAMPS

IndustryMarketing / Advertising

RegionUS

ScaleSME

ChannelOnline

🔧 Workflow

Every day, receive customer complaint text from emails or chats forwarded by sellers. The system first uses a large model to determine whether the type is returns/exchanges or order inquiry, then generates a resolution reply with the brand's tone and sends it automatically. Humans only review complex cases and tickets where the refund amount exceeds the threshold. Inputs are customer complaint messages exported from the seller's store email or chat backend; outputs are ticket records that have been automatically replied to and a high-risk list requiring human intervention. The entire process runs once a day and compiles a weekly resolution-rate report submitted to the seller.

🛠 Setup Requirements

No need to train your own model; use Zapier or Make to connect Gmail and the OpenAI API, prepare 10 to 20 reply templates for returns/exchanges and order inquiries, and you can get it running in 1 to 2 days. The technical barrier is being able to write basic prompts and set up automation workflows; no-code tools are sufficient. If you have Python skills, you can further use the official API to reduce costs.

🧰 Toolchain

🔧 Zapier
🔧 OpenAI GPT model
🔧 Gmail
🔧 Google Sheets

💰 Revenue

① Automated email and chat resolution for small e-commerce sellers (replicating the poster's income): Sellers pay a monthly service fee per store, 500 to 2,000 yuan per store × 5 to 10 stores = about 8,000 to 20,000 yuan in monthly income, accounting for about 100% of monthly income (converted from data in the card, seller self-reported, not independently verified); ② Additional processing fee based on ticket volume: after exceeding the base volume, 1 to 3 yuan per ticket; during peak periods, a single store's monthly fee can reach more than 3,000 yuan, which is incremental beyond ①. The share of this increment is not broken out (seller self-reported, verification lacking); ③ Company-level reference: Netomi's email and chat auto-resolution subscription for enterprise customers; its ARR and unit price are not disclosed in verifiable sources, and the proportion of total revenue from this line is not disclosed, so it cannot be converted; ④ Opportunity item—turn returns/exchanges and order inquiry scenarios into a universal template package for small sellers and charge by template subscription; there is no data yet on how much it can earn.

💸 Cost

Zapier subscription is about 160 RMB per month; GPT API is billed by processing volume at about 200 to 500 RMB per month. If you use Make instead of Zapier, tool costs can be lowered to within 100 RMB per month. API costs grow linearly with ticket volume, but the average cost per ticket is usually less than 1 yuan.

⏱ Time Investment

8 to 12 hours per week, mainly spent reviewing high-risk tickets and adjusting reply templates. Initial setup and template preparation require about 8 hours over two consecutive days; afterward, weekly maintenance time can be compressed to within 6 hours.

🚀 Getting Started

Step 1: Post an e-commerce customer service automation service on Fiverr or Upwork, priced from 500 yuan per store per month. First use a free Zapier account plus GPT to handle 3 test stores and collect screenshots as case studies. Step 2: Compile the before-and-after comparison of the test stores into a one-page case study, using specific numbers to show that refund processing time was shortened from several hours to a few minutes, then pitch it to small sellers on Etsy or Shopee.

🔑 Keys to Success

✅ Only take on highly repetitive scenarios such as returns/exchanges and order inquiries; do not touch complex customer complaints
✅ Keep a manual review entry point for every automated reply to avoid accidental refunds
✅ Use the seller's historical tone as few-shot examples so replies do not sound like a robot
✅ Charge by resolved ticket volume rather than hours worked, so sellers directly perceive the value

⚠️ 风险

⚠️ Automated email sending may trigger spam filters, so you need to warm up the account gradually
⚠️ If refund automation goes wrong, it can directly cause financial loss for merchants; a refund amount cap with manual approval must be set
⚠️ Platform dependency risk: policy changes to Gmail or the OpenAI API may interrupt the workflow
⚠️ Customer service messages involve real buyer privacy; a data breach would lead to compliance and trust losses

📌 Real Cases

📌 Netomi itself serves enterprise customers such as Western Union, handles 2.1 billion annual interactions, and completed a $110 million funding round, validating that automated email and chat resolution works at the enterprise level; individuals can apply this logic to small-seller scenarios and charge per ticket.
📌 An individual used Zapier plus GPT to build a returns/exchanges auto-reply system for an Etsy seller, earning 500 to 800 yuan per store per month; serving 6 stores at the same time yields about 4,000 yuan per month. Source: screenshots of listed prices on similar Fiverr service pages.
📌 A Southeast Asian Shopee seller connected order inquiry auto-replies to WhatsApp, handling about 300 inquiries per store per month, billable at $0.10 to $0.20 per inquiry. Source: publicly available service descriptions in Upwork freelancer order records.

Sources