A Call for Talent from the CVTE AI Team
2026-04-08
ntroduction
Two years ago, when the AI wave first emerged and most people remained on the sidelines, one of our technical colleagues stepped out of his comfort zone and built the Group’s AI application implementation team from scratch.
Starting alone with only an AI "hammer" seeking suitable "nails" across the business, the team has now grown to 6 members, with products covering core businesses including MAXHUB, Seewo, supply chain and overseas markets. Over the past two years, they have boosted the AI adoption rate of data development to over 50% (peaking at 90%), and completely solved the BI operation and maintenance challenge that had plagued the team for four years with ChatBI.
This is not a solo hero story, but a real possibility nurtured by a group of innovators and changemakers on CVTE’s platform. Today we send out this call: if you are equally passionate about AI and aspire to leverage technology to multiply efficiency tenfold, welcome to join our team that redefines work with AI.
Below is the colleague’s firsthand account of the two-year journey — filled with excitement, confusion, setbacks, and remarkable breakthroughs.
Reading Guide
From 0 to 1: One pioneer charting the course, three building the demo, six launching the product — How the ChatBI team took root amid trials and tribulations.
Days Ignited by AI: Burning tokens and a group of relentless innovators.
What We Have Achieved: A full AI workflow spanning data development to data consumption, backed by solid real-world figures.
Our Transformation Over Two Years: From Java backend developers to strategic product thinkers engaging with senior management.
Heartfelt Reflections: On the possibilities, uncertainties and choices in the AI era.
Preface
In the first few months, we faced constant rejections with occasional sparks of inspiration, yet mostly lingered in confusion: Could AI truly be implemented and applied within an enterprise?
Two years on, I can say with certainty: Absolutely, and its impact far exceeds our imagination!
What I want to share most is this: my coding output, technical growth, personal breakthroughs and product deliverables over these two years have surpassed the sum of the past decade. AI has given ordinary people like me a lever, empowering us to achieve far more than we ever expected. It is a pleasant surprise, and also a source of profound awe.
I write this article not to boast, but to record every moment of excitement, confusion, failure and breakthrough over the past two years. If you are hesitating to embrace AI, or feel lonely on the journey of AI implementation, I hope this story can give you strength.
I. From 0 to 1: An AI Adventure of a Data Team
Our story began in February 2024.
Back then, we just started exploring AI application implementation with no teammates and only a vague vision: leveraging large language models for data-related scenarios. Those early months were the toughest. Armed with AI as our tool, we searched relentlessly for applicable business scenarios and visited various departments like sales representatives, only to return empty-handed most of the time. It was never a lack of collaboration from colleagues, but rather uncertainty about which scenarios were truly viable.
Anxiety loomed large — we constantly doubted whether this direction could succeed. I even dreamed of poring over business pain point documents of every department, fearing we might miss a feasible implementation opportunity.
The turning point came with the monthly expense verification in the Finance Department. Expense reimbursement documents exported from dozens of subsidiaries featured inconsistent formats and mismatched account codes. Three staff spent an entire week cross-checking, yet the company audit still identified two unrecognized cross-period expenses. Staring at piles of printed documents on their desks, I suddenly realized: this was exactly the "nail" we were looking for.
I spent two days building an automatic aggregation and cross-verification workflow with large language models, integrated directly into the expense verification process. The workload that once took three people a whole week now finishes in just two hours, with automatic identification of abnormal account classification and cross-period financial risks. When the finance manager came to me with the verification report, the weight on my mind finally lifted.
Subsequently, more departments reached out proactively. The Supply Chain team wanted AI to automatically aggregate production scheduling data and trigger anomaly alerts; the Marketing team hoped AI could analyze multi-channel campaign performance. It was also during this time that I recruited our first team member.
As demands piled up, we spotted a common pain point across nearly all departments: needing data to answer business questions. Instead of running customized data analysis for every demand repeatedly, why not turn this capability into a product? One that allows business users to ask questions in natural language, with AI handling data queries automatically.
This marked the birth of ChatBI.
In October 2024, we launched the first demo based on Tencent’s open-source SuperSonic, which was warmly received by pilot users in the Finance Department.
In February 2025, influenced by restructuring within the open-source community team, we struggled to meet user demands and faced user attrition.
In May 2025, we switched our technical route and officially launched ChatBI 1.0 rebuilt on Dify Workflow. Though not fully meeting our ultimate vision, it achieved stable enterprise implementation, covering 80% of target users. Sincere praise from users on various occasions greatly inspired our team.
Later, the team expanded into three business lines: external product offerings, internal data knowledge base, and AI application development. I led part of the team back to the IT Data Team, launching a large-scale initiative to reshape data workflows with AI, covering CVTE’s core scenarios including supply chain, marketing and finance.
There has been no smooth sailing along the way; we forged ahead entirely through trial and error. This journey from 0 to 1 has forged a team unafraid of failure and bold enough to restart from scratch.
As pinned in our team project workspace:
Difficult as the task is, perseverance leads to success; far as the journey is, progress gets us there.
II. Days Ignited by AI
If asked what was the most extraordinary part of these two years, I would say: we were completely ignited by AI.
It was like being immersed in a captivating game. For an entire month, I stayed in a state of high excitement, tossing and turning at night with constant ideas swirling in my mind. I often got up in the early hours to delve into coding and verify technical feasibility, constantly driven to develop more functions and validate new ideas.
Development tasks that once took two to three days can now be fully completed and delivered within one or two hours. I spend at least half of my day interacting with AI tools, consuming tokens at an astonishing rate — even multiple premium AI accounts could not keep up with my usage.
Here are some examples: Completing dimension value management based on graph databases in 2 hours (end-to-end from conceptual design, interactive interface development, front-end and back-end integration to code submission); Upgrading JDK from 1.8 to 17 in 3 hours; Mastering the source code and architecture of open-source projects WrenAI and Wren-engine in 3 days — work that would have taken 3 months previously.
Over these two years, I have written over 1 million lines of code, exceeding my output of the past decade, with 80% of the work assisted by AI. I have launched more than a dozen major Git projects, surpassing my record of the previous ten years. My spending on AI tokens alone has outstripped all my software subscription costs over the past decade.
During this period, I applied for a top-spec MacBook Pro M5 Max from the company, with memory larger than my phone’s storage. With high-end hardware and ample token supply, watching real-time logs scroll up the terminal elevated the entire AI collaboration experience. This efficiency leap made me realize profoundly: AI is not merely an auxiliary tool, but a capability amplifier that condenses a decade’s work into two years. Tasks once deemed impossible are now routine breakthroughs.
The rest of the team experienced the same transformation. Empowered by AI, everyone broke their capability limits and unleashed remarkable potential.
Yan thoroughly mastered the underlying architecture of BI systems and resolved the four-year-old data challenge with AI. His expertise now supports MAXHUB supply chain analysis and educational business reporting, cutting 2-day data repair work down to just 15 minutes.
Xianggu secured second place in the Greater Bay Area AI Challenge.
Hao mapped out product strategies during train rides, drawing inspiration from classic strategic thinking to propose a vertical-horizontal development framework: deepening professionalism vertically to retain users, and exploring innovation horizontally to attract new users.
Yangyang and Xian played the role of critical challengers, constantly questioning product functions and design: Why this approach? What real business value does it deliver?
Everyone participates in product positioning and functional iteration — every team member acts as a product manager and architect. Debates, discussions, persuasion and compromise have become our daily routine.
When a group of people with insatiable curiosity for AI come together, incredible chemical reactions are inevitable.
III. What We Have Achieved
ChatBI (Intelligent Data Query)
Users query business data in natural language; AI automatically generates SQL, runs queries and delivers analytical conclusions. With an intent recognition accuracy of over 97%, it covers CVTE’s core business lines.
DWStore (Data Knowledge Base)
We have integrated all of CVTE’s data assets into one unified platform, compatible with all mainstream Agents. Data developers can intelligently explore data assets, check indicator standards and analyze historical requirements — eliminating repetitive SQL workloads entirely.
The 24-member data development team has achieved an average efficiency boost of 50% (peaking at 90%). Data demands that once took 2 days now finish in half a day.
As one colleague put it: "80% of my SQL code is generated by this tool." The saved time is fully dedicated to complex business modeling and architectural design.
BI + AI Assistant
Operation and maintenance has always been labor-intensive work, yet our solution has resolved nearly all pain points in BI O&M.
Features include automatic permission activation and intelligent data anomaly diagnosis, bringing qualitative improvements to data delivery efficiency and user experience.
Freed from SQL coding and repetitive O&M work, data professionals can focus on in-depth data insights, boosting job satisfaction while empowering business teams:
- Employees no longer need to consult others for data; instant results are generated via AI input. 30% of low-value repetitive work in sales management roles is eliminated, allowing staff to focus on meaningful work and gain greater career fulfillment.
- Monthly business reports are released 3 days in advance, enabling management to identify issues and make timely decisions earlier.
We have built a full end-to-end AI-driven data workflow, covering data development (DWStore+AI), report configuration & O&M (BI+AI Plugin), and data consumption (ChatBI).
This is not just an architectural diagram on a presentation slide — it is a practical system running in production, used daily by employees across the company.
IV. Our Transformation Over Two Years
Beyond product iteration and technological expansion, AI has profoundly reshaped and empowered every individual on the team.
Two years ago, we were purely Java backend developers and data analysts, accustomed to coding silently, hesitant to communicate with business departments, let alone engage with senior management. Building ChatBI forced us to step outside our comfort zone: we had to understand sales logic and educational business operations, persuade users, secure resources and make critical product decisions.
We jointly launched an official WeChat account, growing it from scratch to over 10,000 followers. This journey honed our communication and critical thinking skills, built internal and external influence, and led to multiple industry sharing invitations. Everyone’s soft skills evolved dramatically: we now think from a product perspective, develop professional acumen, and can systematically analyze and solve complex business problems when engaging with senior leaders.
Our understanding of team building has also undergone a profound shift. ChatBI is more than a tool or product; it serves as a carrier for nurturing AI talents, building an AI-driven workplace, and revolutionizing traditional data workflows. We therefore spent 7 months on campus recruitment, internally named the "Friend Hunt Initiative" or "Long March". Why invest so much time? Because we firmly believe: Finding the right people matters more than writing perfect code.
We have summarized a core ecosystem model: Models, Engineering, Talents, Environment form an interconnected value chain. The strength of models defines capability limits; refined engineering enhances user experience; professional talents deliver high-quality outcomes. When models and engineering are fully ready, AI productivity overflows exponentially — yet most individuals and organizations are not prepared to embrace this transformation.
This is why we prioritize talent recruitment above all else. We are not looking for mere executors, but innovators and changemakers.
V. Heartfelt Reflections
After two years of AI implementation, I want to share some genuine insights with you.
AI productivity has already overflowed, yet most organizations remain unprepared.
In 2026, AI Agents will be widely deployed across enterprise scenarios, making human-AI collaboration the new normal — there is no turning back. This presents a once-in-a-lifetime window: the earlier you embrace AI, the faster you build competitive advantages.
Future core competitiveness lies not only in technical expertise, but in imagination, questioning ability and critical thinking beyond technology.
- Imagination empowers you to create innovative products with AI-amplified capabilities.
- Questioning ability determines how effectively you leverage AI to find solutions.
- Critical thinking keeps you clear amid information overload — easily acquired answers often come with massive noise.
Don’t fear being "replaced".
Our team has a motto: See opportunities as innovators amid the anxiety of being disrupted. Keep moving forward, solve unsolved challenges, explore uncharted fields, and simplify tedious, inefficient and unreasonable workflows. Dwelling on hesitation solves nothing; taking a step forward brings clarity.
The era of the super individual has truly arrived.
In the future, everyone will be a problem-solver, value creator and ecosystem builder. It will become increasingly common for one person to complete full-cycle work including research, design, development and marketing. The entry barrier keeps lowering, while the career ceiling keeps rising.
To Enterprise Managers
AI is far more than an auxiliary tool for incremental improvement; it is core infrastructure reshaping work models. Our practical experience fully proves that building an AI-ready workplace with deep business and data understanding can significantly accelerate decision-making speed, accuracy and comprehensiveness — and this path is entirely replicable.
Our real-world results speak for themselves: 80% core user coverage for data consumption, 50% average efficiency boost for the 24-person data team, and complete resolution of the four-year BI O&M challenge. These are tangible production results, not empty marketing claims.
Technical advancement is the foundation, yet organizational determination is the key to successful AI implementation. Enterprises need to advance AI initiatives with patience, confidence and resolve, guided by clear management indicators, to fully unlock AI’s business value.
To Young Job Seekers
Talent speaks louder than background. We welcome you to join us if you possess the following qualities:
- Curious Innovator: Eager to test new tools and delve into emerging technologies. This innate curiosity is more valuable than any existing skill.
- Bold Explorer: Unafraid of failure or restarting from scratch. Our ChatBI solution has undergone failures, reconstruction and restructuring — every reset has made our team stronger.
- Action-oriented Thinker: Able to conceive AI-driven improvements and turn ideas into tangible results, even for small daily tasks.
If you feel lost and unaccomplished right now, don’t let anxiety drain your passion and resolve. Calm down and focus on what you need to do. It is never too late to start. Begin by trying an AI tool or completing a small project, accumulate experience through practice, polish your capabilities, and gradually build your core competitiveness.
To Casual Readers
I hope this article can be a glimmer of light in your low moments, giving you the courage to restart from scratch. If it inspires you to explore AI, it will be my greatest encouragement.
Epilogue
Some ask us: Is working on AI in a non-internet enterprise rewarding?
Our answer: A hundred times more rewarding than imagined.
At CVTE, AI development is never theoretical. We engage with real business scenarios, tackle core user pain points, and leverage massive authentic data. There is unparalleled satisfaction in using AI to solve long-standing team challenges, implementing AI into core links such as supply chain scheduling and R&D decision-making, and witnessing tangible efficiency improvements — turning AI from abstract concepts into real business value. No routine technical training can compare to this sense of creation.
When we first embarked on Data+AI product development two years ago, we never imagined where we would be today. Two years later, we still cannot predict our future path. But one thing is certain: opportunities abound on this journey; what we lack most are creative, forward-thinking talents.
I’d like to share a quote for all explorers on the AI implementation journey:
Don’t wait for perfect models. Dive into the field, embrace the limitations of current models, and actively test next-generation iterations. Only then will you realize you can finally turn your visions into reality — with just a little more advancement in AI capabilities.
If you feel even a slight urge to give it a try after reading this, don’t hesitate — let’s connect!
We are launching a special campus recruitment program for AI talents. Scan the QR code below to submit your resume directly, and join us to explore the amazing new way of working with AI!
The ChatBI Team — redefining data work with AI — is waiting for you to join.
We provide access to cutting-edge large language models with full priority usage. Our team includes top 20 core developers in the Dify open-source community and a host of AI experts. Dozens of AI scenarios have been fully implemented across supply chain, finance, marketing and other fields. This is a truly AI-First workplace: we are heavy AI users ourselves, and we understand your passion for artificial intelligence — because we share the same drive.
return
share
