Português
EnglishBuild Your Own CRM AI Assistant in Half an Hour
QwenWork Hands-On Tutorial
The CRM system you invested heavily in ends up being one-size-fits-all—
Three months after launch, after round after round of training, only a few people are still using it consistently: the IT administrator maintaining the system, the Sales team entering leads, and the manager checking data at month-end.
Sales reps say, "After a full day out in the field, I have to fill in follow-up notes in the evening. It's not just notes—it feels like writing a memoir."
The manager says, "Half the fields in the spreadsheet are empty. How can I make decisions—or exercise my imagination—with this?"
The IT team says, "The requested features are scheduled for next quarter. For now, everyone will have to make do."
The problem isn't that people aren't trying hard enough; it's that we started off on the wrong track: we're forcing people to adapt to the tool.
Change Your Approach—Let the Tool Serve You
The issue with traditional CRMs is that they're passive: if you don't push them, they won't move.
In the AI era, what we need isn't a more perfect CRM—it's an AI assistant that proactively drives work forward—
It works automatically: every morning at 9 AM, a daily report is sent to DingTalk, prioritizing which clients to follow up with, which deals to push, and which numbers to chase—all arranged by urgency, no extra effort needed.
It organizes automatically: after a call, just say, "The client said they'll pay after the holiday," and it expands that into a complete follow-up record and archives it; contracts and proposals are saved too. Next time you draft a new proposal, it pulls up old documents and past interactions with that client as references.
It analyzes automatically: turn a spreadsheet into a dashboard with one click; quarterly reports stop being just a list of numbers and start telling the story of the current situation, pinpointing problems, and offering solutions.
Today, let's use Qwen Office + DingTalk AI Tables + DingTalk to build an AI assistant that automatically helps you drive sales, so your CRM stops sitting idle—
First, Let’s See the Results
No need to take our word for it—see how this AI assistant actually works:
Interaction Layer – DingTalk App: View Data Anytime, Anywhere
<‹ Swipe left to view
Data Layer – DingTalk AI Tables: Data-Driven Access, Permission-Based
<‹ Swipe left to view
Execution Layer – Qwen Office: Follow Your Commands, Visualized Output
<‹ Swipe left to view
Each layer plays its role, and together they form a complete AI assistant.
Four Steps to Build It Yourself
Step 1: Define the Role
When onboarding an employee, the first step is to set the context: what business are you in, and what tasks do you want it to handle?
If the background is clear, it knows where to focus.
Open Qwen Office and feed it this prompt directly:
Prompt (copy and paste): Hello, I'm the sales manager at a medical equipment company. I want you to be my sales operations assistant, helping me follow up with clients, advance team projects, and remind me of pending tasks. You're goal-oriented and should work closely with me. We mainly sell XX products, focusing on customer payments and contract renewals.
There are two key points here:
Always specify the industry. Customer logic varies widely across sectors—healthcare focuses on payments and renewals, government work centers on process milestones, and e-commerce tracks repeat purchases. Be specific so it knows where to direct its efforts.
Overcommunicate the background. What products you sell, how many team members you have, which metrics matter—feed it all. The more information it has, the better its judgment will align with your business needs.
This context gets stored in Qwen Office's long-term memory ("Settings" → "Memory"), and it carries this context into every conversation going forward.
Step 2: Set Up the Data
You don't have to build the tables yourself—just ask AI to do it:
Prompt (copy and paste): Please create a DingTalk knowledge base called "AICRM," containing an AI table named "Customer Management" with four sheets: Customers, Team, Follow-Up Records, and Payments. Also create a folder for contracts, proposals, quotes, and other documents.
If you haven't connected DingTalk yet, enable the DingTalk connector in Qwen Office first.
Once the tables are ready, have it read through your database:
Prompt (copy and paste): Scan the AICRM knowledge base and list all the tables along with their columns.
Here are two tips:
One follow-up records sheet is enough—no need to categorize manually; that's the AI's job. Just say, "Record a follow-up," and it handles filling in the form, linking it to the right client, and updating the status itself.
Contracts, proposals, quotes, and even meeting minutes recorded by the AI—all can go into the knowledge base. When drafting a new proposal for a client, it can pull up past proposals and communication notes for reference. Once the proposal is generated, download it as Word or PDF with one click and send it straight to the client.
Step 3: Set Up the Workflow
This is the critical step that turns passivity into proactivity.
Run a test first:
Prompt (copy and paste): Identify which clients need follow-up today and which payments are overdue. Generate a list with suggested talking points for each item.
If the results are satisfactory, upgrade it to a scheduled task:
Prompt (copy and paste): Make this workflow a daily 9 AM task and send me the output. If each item has an assigned owner, also send the relevant details to their DingTalk inbox.
From now on, every morning your DingTalk will receive a sales briefing: high-priority clients, today's to-dos, overdue payments—each item comes with suggested scripts you can use directly.
Daily follow-ups become effortless too. After meeting a client, simply tell the AI:
I just sent xx client a payment reminder. They said they'll pay tomorrow—great news!
It automatically writes the follow-up note, links it to the client, updates the payment status, and even reminds you: verbal promises aren't binding—don't forget to confirm receipt tomorrow.
Step 4: Optimize the Experience
The first version of your AI assistant will likely just be a "data mover." For example, asking it to generate a quarterly report might produce nothing more than a collage of dashboard screenshots—a page after page of raw numbers.
Don't settle—ask it to revise:
Prompt (copy and paste): This content is too dry and lacks insight. It's not a proper report. Start with a single sentence summarizing the current situation, then back it up with data, and finish with actionable steps and recommendations.
The revised output looks completely different: a concise opening line—"Customers are growing, but cash flow hasn't kept pace"—followed by a breakdown of 837,000 opportunities, a 22% payment rate, and a focus on which customer segments are behind schedule, culminating in clear next steps.
AI-generated results are 80% about guidance and 100% about iteration. If something isn't working, explain it plainly and ask for changes. Each revision makes it better.
Advanced Features
Once the basic setup is running smoothly, you can expand further:
Create a sales dashboard: Have Qwen Office convert the data in the AI table into a live dashboard—showing customer distribution, sales pipeline, payment progress, and team workload. Project the dashboard onto a large screen for departmental reviews and morning meetings.
Automated group chat follow-up: If you have dedicated DingTalk groups with clients, Qwen Office can periodically extract key insights from conversations and surface follow-up tasks, reducing manual data entry. Through connectors, it can even read chats from Feishu, WeChat Work, WhatsApp, and WeChat.
Integrate with existing CRMs: If you already have a CRM system, ask your IT team to expose its API as an MCP endpoint for Qwen Office, allowing it to operate on top of your current system.
Set role-based data permissions: When multiple users share the same dataset, configure advanced access controls in AI Tables—for example, give Sales reps access only to their own accounts and managers access to their teams' data.
Conclusion
AI Tables are like a versatile building block—you can construct almost any lightweight business system with them.
But being able to build is one thing; building well is another—many users find the learning curve steep and struggle to get started.
Now, tasks like creating tables, pulling data, and automating workflows can all be handed off to Qwen Office. Just give it simple instructions.
DingTalk manages accounts and outreach, AI Tables handle data and permissions, and Qwen Office takes care of execution.
With this powerful trio, you can build most business systems yourself.
What would you like us to demonstrate next? Share your suggestions in the comments!
· · ··· · ·
A deep-dive column to make Qwen Office even better—become an Agent expert
DomTech is DingTalk's official service provider in Macau, dedicated to serving customers with DingTalk solutions. If you'd like to learn more about DingTalk platform applications, contact our online support, call +852 95970612, or email [email protected]. Our skilled development and operations team brings extensive market experience to deliver professional DingTalk solutions and services!