Customer case studies · Deployed
AI won't replace you. Someone using AI will.
Learn the real AI hard skills, and build strength that holds upThey started with the course, then brought AI into the business
These are results reported by our learners and enterprise customers. Every company has different processes and data. The examples show which parts of a workflow AI can take over; they are not promises of results.
The case study details and figures were supplied by customers and published with their permission. Company names have been withheld at their request.
After a month in the practicum, they handed the entire immigration workflow to AI
Step 1 · Hands-on practicum
During the practicum, they first established the customer acquisition process with guidance from their instructor:
What was deployed
Real-time monitoring of local forums to discover and capture leads automatically
Automatic creation and scheduled publishing of Xiaohongshu image-and-text posts
One-click conversion of posts into videos, followed by automatic publishing on Xiaohongshu for targeted promotion
An AI Agent that creates and sends high-quality promotional materials on a schedule
Customer-reported results
Leads rose 70%, and signed clients rose 30%.
Step 2 · AI enterprise deployment service
After seeing the benefits, they signed up for the AI enterprise deployment service. We helped them map the entire immigration process and turn it into one automated workflow:
What was deployed
A connected process covering consultations, file opening and preliminary review before document submission
Automatic monitoring and alerts for system updates on the immigration authority's website
One AI Agent monitoring 120 customer cases at the same time
Customer-reported results
They saved substantial staffing costs and resources. More importantly, the customer experience improved across the entire process, while errors in manual steps fell noticeably.
Before funds go out, AI scans the customer's background
A risk-control process they built after completing the enterprise deployment practicum:
What was deployed
Automatic AI screening of loan applicants' backgrounds
Early identification of document and background risks, with a risk assessment
Detection of characteristic patterns associated with professional lending fraud
Intelligent tracking and management of bad debt, with automatic alerts
Customer-reported results
The risks of bad debt and lending fraud fell substantially.
New demand on a forum is captured as soon as it appears
Built using TopEdu's AI operating foundation and marketing automation service:
What was deployed
Real-time monitoring and capture across all online forums, so potential leads surface immediately
Automatic follow-up and responses so opportunities are not missed
An AI SEO foundation that continually produces searchable content
Xiaohongshu XOR compliance check + automatic posting to Instagram and Facebook
Customer-reported results
Marketing shifted from relying on staff monitoring to intelligent, precisely targeted delivery.
Inventory and deliveries, with AI doing the arithmetic
AI was introduced to optimise inventory and delivery management:
What was deployed
Accurate inventory demand forecasting to reduce overstocking and shortages
Optimised delivery scheduling and dispatch
Customer-reported results
The customer experience improved noticeably, and inventory forecasts became more accurate.
Content that once required 3 people now needs 1
After the practicum, they automated the entire Xiaohongshu content workflow:
What was deployed
AI automatically handles posting, compliance checks and post optimisation
Real case images are supplied to AI, which automatically creates high-quality posts in the industry's most popular image-and-text styles
Content is published automatically on a schedule to reach users in mainland China with precision
Customer-reported results
The staffing needed to keep the content going fell from 3 people to 1, and they are acquiring more customers, of better quality.
Quotes, drawings and production scheduling, with AI handling the repetitive work first
Sales and production each had their own bottleneck, so we started by handing off the most repetitive work on both sides:
What was deployed
Automatically parses customer enquiries and prepares an initial quote based on the model and materials
Automatically checks drawings against specifications and flags dimensions that cannot be manufactured
Forecasts production scheduling and lead time, showing whether an order can be accepted as soon as it arrives
Automatically answers common dealer questions without taking up the sales team's time
Customer-reported results
Quotes now go out the same day instead of waiting for a salesperson to become available, and batches requiring rework due to specification errors have fallen noticeably.
Files, filing deadlines and first drafts, handled by an assistant that never misses a date
Missing a deadline was the firm's biggest worry, so that is the workflow we started with:
What was deployed
Automatically files case materials and extracts key facts and points
Monitors deadlines and filing schedules, with automatic alerts as they approach
AI prepares first drafts of standard documents for a lawyer to review and finalise
Sorts consultation emails automatically; AI drafts the reply to common questions and a lawyer reviews it before it goes out
Customer-reported results
The risk of missing a deadline has fallen sharply, and lawyers get their time back for the work that genuinely needs judgement.
From site measurement to quote, several days reduced to one
Slow quotes lose business, so we began by streamlining the proposal stage:
What was deployed
Uses AI to interpret customer requirements and site photos, then prepares an initial proposal and materials list
Automatically generates a quote based on materials and labour
Schedules projects and allocates workers, with automatic conflict alerts
Automatically shares construction updates with customers, so tradespeople do not have to send WeChat messages themselves
Customer-reported results
The quote cycle has gone from several days to the same day, and customer follow-ups no longer depend on someone remembering.
Weather, scheduling and follow-up, all running as one automated flow
Roofing depends on the weather, so both scheduling and customer acquisition need to move with the forecast:
What was deployed
Automatically finds and forwards roofing repair requests from local forums and social platforms
Adjusts construction schedules using weather forecasts and automatically notifies customers
Automatically generates quotes and warranty documents
Follows up automatically after completion and invites customers to leave a review
Customer-reported results
Rainy-season rescheduling no longer means calling every customer individually, and the time from a new lead appearing to receiving a response has shortened noticeably.
One person can manage store content and reviews
Interrupted content updates and negative reviews left overnight are two of the most common gaps in restaurant operations, so AI monitors both:
What was deployed
Automatically creates posts for Xiaohongshu and Instagram, schedules them after an XOR compliance check
Turns a single dish photo into written content, images and short videos
Automatically classifies reviews across platforms and alerts the store manager to negative feedback immediately
Generates campaign copy in batches for holidays and new menu items
Customer-reported results
Store content has moved from irregular updates to daily publishing, and negative reviews are generally handled before the next day.
For the two-hour rush, dispatch goes to whatever computes faster
When delivery capacity and orders fall out of step, complaints follow, so we started with dispatch and exception handling:
What was deployed
Forecasts order peaks and allocates delivery capacity in advance
Automatically optimises delivery routes and batches
Automatically identifies late orders, missing meals and address issues, then alerts the right person
Automatically answers customer enquiries and shares delivery progress
Customer-reported results
Late orders at peak have dropped, and support is no longer buried under “where is my food?”
Listing copy, pricing and guest complaints, handled by AI end to end
Once there were more properties than anyone could watch by hand, the whole operation was automated:
What was deployed
Automatically creates listing content for multiple platforms and publishes it across them after an XOR compliance check
Provides dynamic pricing recommendations based on the season and local events
Automatically sends messages before and after check-in, ensuring guest questions receive a response at any time
Automatically assigns cleaning and maintenance work orders
Flags negative reviews and complaints automatically, then escalates them to the person responsible
Customer-reported results
The number of properties one person can manage has increased noticeably, message coverage no longer has gaps, and vacancy periods are shorter.
Admissions enquiries and parent communication without constant WeChat monitoring
Admissions enquiries peak in the evenings and at weekends, when staff cannot always be available, so AI handles the first response:
What was deployed
Answers admissions enquiries 24/7, covering courses, tuition fees, application materials and key dates
Automatically files parent communications and creates follow-up reminders
Automatically prepares admissions content and parent letters with consistent messaging
Automatically compiles student progress reports, leaving teachers to add only what is needed
Customer-reported results
Enquiry responses have expanded from office hours to round-the-clock coverage, allowing admissions staff to focus on families who genuinely need a conversation.
Which part of your industry can AI take over?
Start with a free needs diagnosis. We review your company's current workflow, identify which steps can be handed to AI and recommend the most cost-effective place to begin.
Enrol in AI practice that actually lands
Scan with WeChat to confirm dates, seats and how to enrol, directly with us.
On a phone? Press and hold the code to recognise it, or screenshot it and pick it from your album in WeChat's scanner.
Questions? Ask the AI assistant at the bottom right, or find contact details on the privacy page