Accepting orders with Manus to run deep research and delivery reports, earning 20k RMB per month solo

💰 Revenue

① Company level (Manus platform itself): Users subscribe monthly; source disclosures indicate annualized revenue of $125 million within just 8 months of launch. The number of paid users was not disclosed (estimated by media, unverified), and the exact proportion of its own revenue is undisclosed. ② Replicator level (task outsourcing charged per order, main income): Clients pay research fees per order—deep research reports 500 to 2000 RMB/order, itinerary planning and data sorting 200 to 500 RMB/order, complex industry analysis 1500+ RMB/order × 15 to 20 orders completed per month = monthly income of about 20,000 RMB, covering almost 100% of total monthly income (calculated from internal data, case study self-report). ③ Early invitation code resale (gray market): Source disclosures show internal test invitation codes were scalped on secondhand platforms for 50,000 to 100,000 RMB; actual transaction volume is unverifiable, and revenue contribution is unclear (media accounts, unverified by anyone). ④ Opportunity point—Revenue sharing by project with tourism and consulting agencies: Commission sharing by project with agencies; the percentage ratio has not been publicly finalized, and revenue contribution is unclear.

💸 Cost

Manus subscription fee is several hundred RMB per month. Token/credit consumption based on task complexity needs to be controlled per order. Combined with promotional fees and membership fees on freelance platforms, the monthly cost is under 1,000 RMB, accounting for less than 10% of revenue.

⏱ Time Investment

About 3 to 4 hours per day, including 1 hour handling client communication and revision requests, 1 to 2 hours monitoring agent execution and making mid-course corrections, and 1 hour doing final fact-checking and formatting delivery. The time when the agent backend automatically runs tasks does not consume human labor.

🔑 Keys to Success

• ✅ Humans must perform the final data and fact-checking; AI outputs delivered directly will lead to complaints, and verification capability is the core barrier of this business.
• ✅ Distill task workflows into template prompts, continuously shortening delivery times for similar order types to create compounding returns. Deeply cultivate templates for one industry at a time.
• ✅ Control per-order credit consumption by estimating the number of steps before running tasks, avoiding excessive subscription quota burn on a single order that leads to losses.
• ✅ Case study recordings are the lowest-cost customer acquisition asset; continuously showcasing real delivery processes in the public domain is more persuasive than any advertising copy.

⚠️ 风险

• ⚠️ Platform pricing, credit quotas, and feature strategies may change at any time, directly impacting the cost structure, requiring a contingency plan to switch to backup tools.
• ⚠️ Rapid influx of homogenized competitors in general-purpose agents, with similar products like Genspark offering free or low-cost alternatives to grab customers, potentially compressing the pricing power of pure proxy-running services.
• ⚠️ Varying customer acceptance of AI-generated content. If deliverables are discovered to be directly produced by AI without verification, it may lead to refunds and negative reviews, damaging the reputation of the order-taking account.
• ⚠️ Confidentiality and compliance risks exist when dealing with internal data provided by clients; data usage boundaries must be clarified before accepting orders to avoid disputes.