Using AI Agents to provide local merchant short-video customer acquisition services, generating a single-person monthly revenue of 50,000 RMB
Workflow: Every morning, N8N or Coze automatically pulls the previous day's store foot-traffic data, group-buying redemption data,
Key Fields
FIELD STAMPS🔧 Workflow
Every morning, N8N or Coze automatically pulls the previous day's store foot-traffic data, group-buying redemption data, and user comments as the day's material inputs. DeepSeek invokes preset 80+ script templates based on store categories to generate talking-head copy. Jimeng AI generates corresponding scene images, Kling synthesizes 15-30 second store-exploration short videos, and the system automatically applies preset Jianying templates before scheduled publishing to the stores' Douyin and Xiaohongshu matrix accounts. When users comment, consult, or leave private message leads, WeChat Work automatically pushes preset package quotation pitches. Human intervention is only needed daily to export leads, follow up with high-intent customers, and review the next day's script direction. 95% of the execution steps require no manual intervention, producing 8-15 videos and handling 20-80 lead inquiries per day.
🛠 Setup Requirements
Requires basic setup skills in low-code workflow tools like N8N or Coze, and the ability to call large model APIs such as DeepSeek and OpenAI. Deep coding foundations are not required, but understanding platform login state management and anti-ban rules is necessary. Initial workflow setup takes 2-3 weeks, and running a single-store test requires 1-2 months of tuning. During this period, prepare 2-3 test phone numbers, a Jianying Pro membership, Jimeng AI generation quotas, and a WeChat Work account with private domain functions. It is recommended to manually help 1 store run the full workflow for 1 month and obtain conversion data before solidifying it into an automated template, avoiding poor adaptability from a direct launch.
🧰 Toolchain
- 🔧 N8N
- 🔧 Coze
- 🔧 DeepSeek
- 🔧 Jimeng AI
- 🔧 Kling
- 🔧 Jianying
- 🔧 WeChat Work
💰 Revenue
① Monthly subscription agency operations for store short-video customer acquisition (main revenue): Merchants pay a monthly operations fee of 2,000-5,000 RMB per store. Steadily operating 10-15 stores yields a monthly revenue of 20,000-75,000 RMB, which can reach 50,000 RMB as claimed in the card notes, accounting for about 100% of monthly revenue (calculated as 10 stores × 5,000 RMB = 50,000 RMB based on merchant self-reporting, pending independent verification); ② Startup period and renewal structure: Revenue drops to 10,000-20,000 RMB during the first 3 months of testing and expansion, with a renewal rate of over 60% after running through the SOP; benchmark cases in the notes claim that serving 5-6 stores simultaneously with a monthly income of 5,000-15,000 RMB is normal (based on case claims, independent verification is lacking, and the proportion of this path is not publicly disclosed); ③ SME AI automated customer acquisition outsourcing (spillover of the same SOP set): Priced at an initial setup fee of $1,000-$2,000 USD + monthly fee of $200-$500 USD (prices taken from platform public disclosures), the number of closed deals cannot be verified, and no public percentage is available; ④ Opportunity item - Chain brand package replication: Allocated cost per store is about 100 RMB/month with a gross profit margin of over 90% (cost and gross profit figures provided by the notes), and scale has not yet been publicly disclosed.
💸 Cost
Average monthly tool subscription and API costs are 800-1200 RMB, including N8N cloud service at about 150 RMB/month, Jianying Pro membership at 148 RMB/month, Jimeng AI generation quota at about 500 RMB/month, and WeChat Work basic functions at about 200 RMB/month. Costs increase slightly as the number of served stores and generated videos increases, with the allocated cost per store being about 100 RMB/month, and gross profit margins reaching over 90%.
⏱ Time Investment
During the first month of setup and testing, 6-8 hours are invested daily. Once the SOP runs stably, only 2 hours of daily maintenance are needed: exporting the previous day's leads in the morning to assign for manual follow-up, scheduling that day's video publishing tasks in the afternoon, reviewing each store's account data and adjusting the next day's script direction in the evening, and centrally expanding new clients on weekends.
🚀 Getting Started
Step 1: Choose 1-2 familiar local tracks (prioritize catering and beauty, as they have high standardization and the strongest demand). Manually produce 3 store-exploration short videos in the same category and provide them for free to 3 potential street-side stores to run for half a month, obtaining initial in-store conversion data. Using the data to secure the first partner store, immediately break down the manual script generation, video production, and lead handling processes into N8N node templates. Once the full automation is verified on the first store, replicate it to other stores in the same track. Gradually expand categories only after mastering a single track, without greedily running multiple tracks in parallel.
🔑 Keys to Success
- ✅ Focus on 1-2 industry tracks to create white-label templates (prioritizing catering/beauty and other highly standardized tracks). Fix the single-track script library at 80+ reusable items, sharing costs to pursue high-margin routes.
- ✅ Humans act as referees. Maintain human intervention during the key customer closing stage, leaving machines to handle only front-end interception and lead collection. Avoid blind full-automation that leads to lost orders, ensuring service quality.
- ✅ Operate strictly within platform anti-cheat rules. Do not use unauthorized plugins or engage in batch traffic disruption. Build independent account matrices for each store to extend account lifecycles.
- ✅ Establish a daily data review closed-loop, iterating script and pitch templates weekly to ensure service effectiveness continuously outperforms merchant self-operation, thereby increasing renewal rates and referral rates.
⚠️ 风险
- ⚠️ AI-generated short videos tend to be formulaic. If stores have extremely high demands for conversion results, lost orders may easily occur. It is necessary to intersperse 1-2 weekly manually shot store-exploration materials to supplement a sense of reality and improve conversion rates.
- ⚠️ Sharing accounts across multiple stores or batch-publishing content easily triggers platform traffic throttling and account bans. It is necessary to build an independent account matrix for each store, strictly adhere to platform publishing rules, and use a private domain lead pool as a backup to prevent service interruption caused by bans.
- ⚠️ Local merchant renewals are heavily influenced by physical business cycles. If stores perform poorly, delayed service fees and early terminations are likely. It is necessary to expand clients across 3-5 different tracks to diversify risks, while setting up tiered pricing (low-cost customer acquisition for new stores in the first month, raising prices to increase ARPU after stabilization).
📌 Real Cases
- 📌 xopcx.com records an independent operator using AI Agents for local life services, generating 50,000 RMB monthly and covering multiple catering and medical aesthetics stores.
- 📌 max.book118.com case study: A single person undertaking a full-package short-video customer acquisition service for local merchants, charging operating fees on a per-store, per-month basis.
- 📌 xtcer.cn practical guide shows that individual operators providing AI-automated customer acquisition services for SMEs can achieve monthly revenues of over $5,000 USD, a model highly compatible with local merchant GTM services.
- https://xopcx.com/articles/article-2026-06-04-yiren-local-life-ai-agent
- https://max.book118.com/html/2026/0615/5313331000013231.shtm
- https://www.xueitceo.com/857.html
- https://cloud.tencent.com/developer/article/2697499
- https://xtcer.cn/posts/94df92b3-ba80-4ee7-bc3b-893ba50d7919
- https://n8ncn.io/workflows/zi-dong-hua-ai-ben-di-shang-hu-tuo-ke-yu-she-mei-ying-xiao