← Sticker Wall AGENT · DETAIL

Using Manus to Help Enterprises Implement General Agent Training and Scenario Sorting, Earning 25,000 RMB per Month as an Individual

Workflow: Every morning, collect 3 to 5 actual task requirements submitted by enterprises via WeChat Work communities and private

AGENT

Key Fields

FIELD STAMPS

IndustryEducation / Knowledge

RegionChina

ScaleSME

ChannelOnline

🔧 Workflow

Every morning, collect 3 to 5 actual task requirements submitted by enterprises via WeChat Work communities and private messages, and rank them by feasibility; in the afternoon, use Manus to successfully run one of the typical tasks on-site, recording the entire process and noting failure points and manual correction locations; subsequently, output a scenario feasibility assessment checklist, prompt template package, and acceptance criteria document to deliver to the client; in the evening, conduct a 30-minute Q&A in the client group and schedule the next workshop based on the day's execution results, completing the entire closed-loop on the same day.

🛠 Setup Requirements

Requires proficiency in using Manus to complete multi-step tasks, an understanding of its division of labor mechanism regarding planning agents, browsing agents, and tool-calling agents, and the ability to write clear task instructions and acceptance criteria; prepare a set of stable demonstration accounts, screen recording tools, and Lark document template libraries; during the cold-start phase, spend 2 to 3 weeks running at least 5 cross-browser sample cases, such as competitor information collection, initial resume screening for recruitment, supplier quotation organization, automated industry weekly report generation, and e-commerce review summarization, forming an on-site demonstrable sample library before starting external charging.

🧰 Toolchain

🔧 Manus
🔧 Lark Documents
🔧 OBS Screen Recorder
🔧 WeChat Work Community

💰 Revenue

① Enterprise monthly accompanying service (main revenue): Enterprises pay 12,000 RMB per month (including 4 workshops plus one set of template packages) × 2 enterprise clients per month = 24,000 RMB monthly revenue, accounting for about 96% of monthly revenue (calculated from card number conversions, case caliber, without independent verification); ② Single walk-in client workshops: Enterprise teams pay per session ranging from 3,000 to 5,000 RMB/session, targeting 10 to 30 people, accounting for about 12%-20% of monthly revenue (calculated based on an estimated monthly income of 25,000 RMB from the card, self-stated by the case, lacking independent verification); ③ Repeated sales of template packages: Delivered along with the monthly package and sold additionally as an add-on; neither the single-set selling price nor the number of additional sets sold has public figures, and the proportion of the total pie is not provided; ④ Opportunity item - enterprise seat-based subscription for prompt and template libraries: Using the source-disclosed parallel execution that allows tasks originally taking 8 sequential hours to finish in 3 hours in the cloud as an efficiency selling point, charged per seat, with neither seat pricing nor revenue volume having reliable data (media estimation, unverified independently), and no proportion provided.

💸 Cost

Manus subscription fees cost several hundred RMB per month, billed according to an official credit system, with redundant quota reserved for heavy demonstration scenarios; screen recording tools and document collaboration tools cost about a hundred RMB, totaling approximately 500 to 800 RMB per month.

⏱ Time Investment

About 4 hours per day, including about 2 hours for task execution and screen recording, 1 hour for client communication, and 1 hour for template organization, totaling about 20 hours per week.

🚀 Getting Started

Step 1: Use your own Manus account to run 5 real enterprise scenarios and record the entire process, organizing them into a one-page public casebook annotated with the labor hours saved for each task; Step 2: Publish the casebook on Zhihu, Jike, and industry-vertical communities for free sharing, attracting the first batch of pilot clients with real demonstrations; Step 3: Deliver the first client order at half price in exchange for public attribution rights, client testimonials, and repeat purchase rights, rolling to acquire clients through word of mouth.

🔑 Keys to Success

✅ Only do demonstrations that have been personally executed, do not exaggerate Manus's capability boundaries, and present failure points truthfully
✅ Accumulate reusable templates by industry; the marginal cost of the second delivery of the same scenario approaches zero
✅ Humans act as the final acceptance judges; all deliverables are manually reviewed before being sent out, without dumping agent outputs directly onto clients
✅ Use screen recordings and before-and-after comparative data to prove value, converting abstract AI capabilities into labor hours saved that enterprises can understand
✅ Client success equals renewal; actively follow up on usage effectiveness and iterate templates within 48 hours after delivery

⚠️ 风险

⚠️ Manus tasks have a failure rate and unstable performance, and complex tasks may require multiple re-runs; manual fallback clauses and delivery scope must be clarified in the contract
⚠️ Platform acquisitions and policy changes may affect domestic access stability, pricing systems, and credit rules, requiring backup tool solutions
⚠️ Intensified homogeneous competition in the training market with low-priced imitators following up quickly means building a moat requires deep industry cases and client testimonials
⚠️ Enterprise data security concerns mean tasks involving internal client data require signing non-disclosure agreements and using desensitized materials for demonstrations

📌 Real Cases

📌 According to Tencent Cloud reports, Manus surged to the top of Weibo's trending list within 24 hours of its release, and invitation codes were scalped for 50,000 to 100,000 RMB on second-hand platforms. The official website crashed multiple times due to a surge in traffic, indicating that the demand for early-stage training, agent execution, and consulting services is real and willingness to pay is strong.
📌 A 2026 evaluation by AI Home pointed out that Manus adopts a multi-sub-agent parallel architecture and multi-model routing, achieving an asynchronous working experience of leaving work while tasks are running, which is widely used by operations and research positions for long-task processing. This is precisely the core usage that enterprise training should teach.
📌 Technical analysis from Zhihu columns shows that Manus works through the collaborative operation of multiple modules such as planning agents, browsing agents, tool-calling agents, and code execution agents, achieving a GAIA benchmark completion rate of 91.4%. This proves its capabilities are sufficient to support enterprise-level scenario teaching rather than being mere conceptual hype.

Sources