Cross-Border E-Commerce Product Sourcing Automation: Compress Half a Day’s Work into 3 Minutes with Hermes Agent + XCrawl

A Real Pain Point

Every day, cross-border e-commerce sellers open the Amazon Movers & Shakers page and start their product sourcing work. They need to:

• Manually browse the Top 100 bestsellers, recording rank changes, prices, and ratings

• Copy product titles and translate them into Chinese

• Search for the same products on 1688 and compare prices

• Calculate profits in Excel: selling price - product cost - FBA fees - commission - first-mile logistics

• Repeat the above steps until finding 3-5 products worth testing

This entire process takes at least 4-6 hours.

What if an AI Agent could automatically complete these steps and deliver a structured product sourcing report in 3 minutes? That’s exactly the real case we’re sharing today.

Practical Case: Full Automation from Amazon Bestsellers to 1688 Suppliers

Step 1: Scrape the Amazon Movers & Shakers List

Goal: Obtain structured data for the Top 10 bestsellers in the Home & Kitchen category

API Call:

POST https://run.xcrawl.com/v1/scrape { “url”: “https://www.amazon.com/gp/movers-and-shakers/home-garden”, “args”: { “extract_product_details”: true } }

Returned Result (real API response):

{ “scrape_id”: “01KQYFTW9VRFECS5NQY47E9T18”, “status”: “completed”, “url”: “https://www.amazon.com/gp/movers-and-shakers/home-garden”, “data”: { “json”: { “products”: [ { “rank”: 1, “product_title”: “Owala Disney Princess FreeSip Insulated Stainless Steel Water Bottle”, “price”: “$27.99 - $74.98”, “rating”: “4.7 out of 5 stars”, “review_count”: “14,232”, “rank_change_percent”: “138%” }, { “rank”: 8, “product_title”: “Innqoo Candle Warmer Lamp with Timer, Dimmable Candle Warmer for Jar Candles”, “price”: “$17.98”, “rating”: “4.6 out of 5 stars”, “review_count”: “2,837”, “rank_change_percent”: “89%” }, { “rank”: 4, “product_title”: “Gifts for Mom Birthday Gifts for Women Preserved Rose Forever Flower in Glass Angel Figurines”, “price”: “$45.95 - $52.99”, “rating”: “4.7 out of 5 stars”, “review_count”: “1,292”, “rank_change_percent”: “126%” } ] }, “metadata”: { “title”: “Amazon.com Movers & Shakers: The biggest gainers in Home & Kitchen”, “status_code”: 200, “final_url”: “https://www.amazon.com/gp/movers-and-shakers/home-garden”, “proxy_location”: “US” }, “credits_used”: 5, “credits_detail”: { “base_cost”: 1, “json_extract_cost”: 4 } }, “started_at”: “2026-05-06T11:12:41Z”, “ended_at”: “2026-05-06T11:13:03Z”, “total_credits_used”: 5 }

Key Highlights:

• ✅ Structured extraction — Automatically identifies product cards and outputs a JSON array

• ✅ Complete metadata — Includes status code, proxy location, execution time

• ✅ Transparent cost — Clearly shows base_cost (1) + json_extract_cost (4) = 5 credits

• ✅ Plug-and-play — No HTML parsing needed, directly usable for LLM or data analysis

Cost: 1 call = 5 credits

Step 2: Target Product Selection — Innqoo Candle Warmer Lamp

From the list, we found #8 Innqoo Candle Warmer Lamp worth investigating:

• Rank increase +89% (from #249 to #8)

• Price $17.98 (mid-range, reasonable profit margin)

• Rating 4.6 stars / 2,837 reviews (well market-validated)

Key Information Extraction:

• Product type: Candle Warmer Lamp

• Core features: Timer, dimmable, compatible with jar candles

• Target audience: Home decor, gift market

Step 3: Search for Suppliers on 1688

API Call:

POST https://run.xcrawl.com/v1/search { “query”: “融蜡灯 蜡烛加热灯 批发 1688” }

Returned Result (real API response):

{ “search_id”: “01KQYG111VZZBAA8Q46NDJ19AW”, “status”: “completed”, “query”: “融蜡灯 蜡烛加热灯 批发 1688”, “data”: { “data”: [ { “position”: 1, “title”: “蜡灯灯批发_阿里巴巴”, “description”: “保温蜡烛加热灯Gu10卤素灯杯融蜡香薰灯用光源石英灯. ¥5.80 成交2笔…”, “url”: “https://s.1688.com/kq/-C0AFB5C6B5C6.html” }, { “position”: 6, “title”: “香精灯批发_阿里巴巴”, “description”: “26新款无火可升降香薰融蜡灯实木香薰精油融烛卧室. ¥58.00 成交0笔…”, “url”: “https://s.1688.com/kq/-CFE3BEABB5C6.html” }, { “position”: 9, “title”: “金属烛灯批发_阿里巴巴”, “description”: “融蜡灯创意简约熔烛灯复古日式香薰灯扩香卧室装饰台灯跨境专供…”, “url”: “https://s.1688.com/kq/-BDF0CAF4D6F2B5C6.html” } ], “success_num”: 10, “credits_used”: 2, “credits_detail”: { “base_cost”: 2, “traffic_cost”: 0 } }, “total_credits_used”: 2 }

Key Findings:

• ✅ Returns 10 relevant results from 1688, covering multiple suppliers

• ✅ Price range: ¥5.80 - ¥58.00 (choose different tiers)

• ✅ Multiple results explicitly marked “Cross-border Supply”, indicating a mature supply chain

• ✅ Obtained indirectly via Search API, avoiding direct scraping detection

Cost: 1 call = 2 credits

Step 4: Profit Calculation

Conclusion: Per-unit profit $3.24. If selling 100 units per month, net profit $324. Although the profit margin is not high, the product is lightweight and small, with quick turnover, suitable for batch operations.

Case 2: Preserved Rose in Glass Dome — High Profit Margin Verification

To verify the method’s universality, let’s run another case:

Amazon Product:

• FIACO Preserved Rose in Glass Dome (#110, increase +126%)

• Price: $45.95 - $52.99

• Rating: 4.7 stars / 1,292 reviews

1688 Supplier:

• Supplier: Yiwu Aida Crafts Co., Ltd.

• Price: ¥26.60 (approx. $3.67)

• Monthly sales: 22,000+ units

Profit Calculation (using mid price $49.47 as example):

Selling Price: $49.47 (100%) Product Cost: $3.67 (7%) FBA Fees: $7.00 (14%) Amazon Commission: $7.42 (15%) International Shipping: $3.00 (6%) ──────────────────────── Net Profit: $28.38 (57%)

Profit Margin by Price:

• At $45.95: Net profit $25.28, margin 55%

• At $52.99: Net profit $32.32, margin 61%

Conclusion: The gift category commands high emotional premiums, with profit margins as high as 55-61%, making it a golden category for cross-border e-commerce.

Pitfalls and Solutions: Real Experience Sharing

Pitfall 1: Amazon Product Detail Page Anti-Scraping

Symptom: Using the Scrape API to directly scrape product detail pages returns 404 or blank pages.

Real Case: When trying to scrape the Innqoo candle warmer lamp detail page, the API returned:

{ “scrape_id”: “01KQYG070VD18SC8KCR2VKYB66”, “status”: “completed”, “url”: “https://www.amazon.com/Innqoo-Candle-Warmer-Lamp-Adjustable/dp/B0D1HHSYVQ”, “data”: { “markdown”: “![Sorry! We couldn’t find that page…]”, “metadata”: { “status_code”: 200, “title”: “Page Not Found” }, “credits_used”: 1 } }

Analysis:

• HTTP status code is 200 (request successful), but content shows a 404 page

• Amazon uses this method to bypass simple status code checks

• Indicates Amazon has strict access control on individual ASIN detail pages

Solution:

• ✅ Use the Movers & Shakers list page as the data source instead

• ✅ List pages have lower anti-scraping intensity and contain core information (rank, price, rating)

• ✅ If details are needed, use the Search API to search by product title for third-party data

Cost Comparison:

• Failed detail page scrape: 1 credit (wasted)

• Successful list page scrape: 5 credits (10+ product data)

• Conclusion: List pages offer better cost-effectiveness

Pitfall 2: Direct Scraping of Alibaba Platforms Detected

Symptom: Using the Scrape API to scrape 1688/AliExpress product lists returns anti-scraping interception pages.

Real Case: When trying to directly scrape an AliExpress product page, it returned:

“Sorry, we have detected unusual traffic from your network.”

Analysis:

• Alibaba platforms (1688/AliExpress/Taobao) have strong anti-scraping systems

• Includes JS challenges, device fingerprinting, behavior analysis

• Even with proxy pools, direct access can still trigger interception

Solution:

• ✅ Use XCrawl Search API to indirectly obtain public data

• ✅ Search API retrieves results via search engine cache, avoiding direct access

• ✅ Although data timeliness is slightly lower (1-2 day delay), it’s sufficient for product sourcing

• ✅ Lower cost (2 credits vs. potential wasted retries)

Cost Comparison:

• Direct scraping failure: 1-3 credits × multiple retries = wasted

• Search API success: 2 credits, one-shot

Cost Analysis: ROI Calculation

Total process cost (based on real API calls):

• Step 1 (Scrape Amazon list): 5 credits base_cost: 1 credit json_extract_cost: 4 credits Gets 10+ product data at once

• Step 2 (Product details): 0 credits (extracted from list)

• Step 3 (Search 1688 suppliers): 2 credits base_cost: 2 credits Returns 10 search results

• Step 4 (Profit calculation): 0 credits (local calculation)

Total: 7 credits per product

Free tier: 1000 credits = can analyze 142 products

Comparison with manual cost:

ROI Analysis:

• Free tier can run 142 product cases

• Assume finding 3 testable products (2% success rate)

• Each product monthly profit $300 (conservative estimate)

• Potential monthly revenue: $900, tool cost: $0

Technical Implementation: Hermes Agent + XCrawl

This case uses Hermes Agent as the AI orchestration engine, leveraging XCrawl’s MCP interface for data collection.

Why this combination:

• Hermes natively supports the MCP protocol, one-click integration with XCrawl

• Built-in workflow orchestration, no need to write HTTP requests or JSON parsing

• Supports multi-step reasoning, automatically handling exceptions and retries

MCP Configuration (one-time, one line of JSON):

{ “mcpServers”: { “xcrawl”: { “url”: “https://mcp.xcrawl.com/{API_KEY}/mcp” } } }

After configuration, the Agent automatically gets 4 tools:

• xcrawl_scrape — Single-page structured extraction

• xcrawl_search — Search engine interface

• xcrawl_crawl — Full-site deep crawling

• xcrawl_map — Site map generation

The entire product sourcing flow requires only 3 tool calls:

Step 1: Scrape Amazon Movers & Shakers

xcrawl_scrape(“https://www.amazon.com/gp/movers-and-shakers/home-garden”)

Step 2: Search 1688 suppliers

xcrawl_search(“融蜡灯 蜡烛加热灯 批发 1688”)

Step 3: LLM automatically compares and calculates profit

→ Agent analyzes both results and outputs a profit comparison table

In practice, it’s even simpler: Just tell the Agent in natural language “Help me scrape Amazon Movers & Shakers, then search for the same products on 1688, and calculate profit.” The Agent will automatically break it down into corresponding tool calls.

Comparison with traditional approach:

• ❌ Manually write curl for HTTP requests

• ❌ Parse JSON responses to extract fields

• ❌ Handle error retry logic

• ✅ MCP approach: zero lines of code, just natural language

Conclusion: Tools Are Cheap, Workflows Are Valuable

This case demonstrates not “how to use XCrawl,” but “how to use an AI Agent to restructure the product sourcing workflow.”

Core Value:

• Data-driven decision making — No longer relying on intuition, but filtering by rank increase, rating, and profit margin

• Scaling validation — From analyzing 5 products a day to 100+

• Reducing trial-and-error costs — Calculate profit margins clearly before placing an order

Open Source and Commercialization:

• The basic version Skill (product sourcing workflow) is open source and available on GitHub

• Advanced features (scheduled monitoring, price alerts, multi-platform comparison) are planned as commercial offerings

Final Thought:

Cross-border e-commerce is essentially an information asymmetry arbitrage, and the role of an AI Agent is to minimize that information asymmetry to the extreme.

Get Started Now: Your Competitors May Already Be Using It

Quick Start → Register on the XCrawl website and get your API Key

Free Trial, No Credit Card Required:

• 🎁 Register to get 1000 credits (can analyze 142 products)

• ⚡ Complete setup in 5 minutes and start product sourcing immediately

• 📊 Your first bestseller might be in today’s Movers & Shakers

⏰ Data Timeliness Reminder

The data for the two products in this case (candle warmer lamp, preserved rose in glass dome) is from May 6, 2026. The Amazon Movers & Shakers list updates hourly, so we recommend:

• Regularly scrape the latest list data

• Monitor seasonal product windows (Mother’s Day, Valentine’s Day, Christmas)

• Build your own product sourcing database and continuously optimize your strategy

💡 Frequently Asked Questions

Q: I don’t know how to code. Can I use this? A: Yes. Configuring Hermes Agent only requires copying and pasting one line of JSON, then you can interact using natural language.

Q: Will this violate platform rules? A: No. XCrawl collects only public data, in compliance with Amazon and 1688 terms of service.

Q: What happens when my free credits run out? A: Paid plans start at $29/month, with an average cost per product analysis under $0.20. Compared to manual labor, the ROI is significant.

Q: Is this method suitable for all categories? A: It works best for standardized categories like gifts, home decor, and small appliances. Customized products or those requiring deep supply chain integration need additional evaluation.

Disclaimer: All data collection in this article is based on public information and complies with platform ToS and data compliance requirements. XCrawl is a paid service. This article does not constitute investment advice. Cross-border e-commerce carries risks; product sourcing should be done with caution.