The Industry's "Question of Car Manufacturing"
Hello everyone, industry peers and partners. I'm Zhang Lei, Head of Process IT at Avatr Technology.
Today, I'd like to explore a question with you: When we talk about "building a smart car," what exactly are we creating? Is it just a combination of metal, chips, and software? Or perhaps a technology exhibition on four wheels?
No—it’s neither. At Avatr, our answer is: We’re creating a reliable mobile lifestyle, unlocking more possibilities for the intelligent era.
Behind these "possibilities," however, lies a critical challenge that cannot be ignored: In an age defined by personalization, rapidly changing demands, fast-paced technological updates, and even chip shortages, how can every intelligent vehicle—from design to delivery—be executed reliably and traceably at each step?
This is precisely what I want to share with you today: How Avatr creates certainty amid industry uncertainty.
As you all know, car manufacturing is a complex and massive undertaking. Modern vehicles have long evolved from "mechanical products" into "systemic products," bringing unprecedented levels of "uncertainty" to the entire industry.
During the birth of intelligent vehicles, the deep integration of mechanical systems, electronics, and software dramatically increases system complexity. Meanwhile, over 20,000 components must be developed across research institutes and R&D centers worldwide, requiring close collaboration between OEMs and multi-tier suppliers. Furthermore, regulatory requirements are becoming increasingly stringent, driving up compliance costs; technology iterates at lightning speed, forcing us to continuously adapt to waves of innovation; user needs are ever-changing—demanding both cutting-edge tech and robust safety—while market dynamics shift rapidly, compelling us to respond flexibly.
In such an “uncertain industry,” even an “uncertain era,” what kind of intelligent mobility do users truly need? I believe Avatr has already answered this question.
Avatr is not a traditional automaker. We are a new intelligent electric vehicle brand. More accurately, we are a technology company dedicated to “creating future-leading mobility experiences through original design and intelligent technology.” We bring together dreamers who “embed user needs into every detail,” committed to exploring human-centered mobility technologies for the future and delivering emotionally intelligent travel experiences.
Our product lineup includes the Avatr 06—a smart luxury sports sedan; the Avatr 07—a smart urban luxury SUV; the Avatr 12—a future-oriented smart luxury sedan; and the Avatr 11—a future-oriented smart luxury SUV. These models are rooted in original futuristic aesthetic design, representing the perfect fusion of technology and art.
Avatr’s intelligent system integrates advanced driver assistance with a HarmonyOS cockpit, supporting multimodal interaction to achieve “what you think is what you get.” Deep integration across powertrain, chassis, and autonomous driving domains enables superior three-dimensional perception. Technological highlights such as the Kunlun Extended-Range Core Technology intelligently decide when to switch between fuel and electricity, balancing energy efficiency and comfort. Even under low battery conditions, it delivers strong performance with smooth and quiet driving experience.
I personally summarize this in four keywords: Future Design, Intelligent Interaction, Comfortable Experience, and Energy Efficiency. So here comes the question: How do we deliver such intelligent mobility experiences to users? What enables the realization of this vision? Or more specifically: What kind of project management do we need to ensure stable delivery of whole-vehicle projects?
Let Processes Speak, Let Data Drive Decisions
"Avatr" means "avatar." At Avatr Technology, we give it deeper meaning—we aim to create another version of you within a parallel world of intelligence, emphasizing "emotional intelligence" and striving to become your “intelligent avatar that understands you.” Then, should our project management also be such an “understanding avatar”? One that understands processes, scenarios, data, and your intentions. An avatar that helps teams execute smoothly and effectively, mirroring the same four core attributes as our products: Future Design, Intelligent Interaction, Comfortable Experience, and Energy Efficiency.
It must be able to: 1) Flexibly configure to adapt to different management scenarios, even accommodating special mechanisms; 2) Leverage AI to solve real problems and generate business value; 3) Enable seamless collaboration and integrated interactions, breaking down departmental silos; 4) Reduce costs, improve efficiency, and shorten delivery cycles.
In short, it allows teams to focus on important work by minimizing distractions, ensuring every vehicle is delivered stably and with high quality into users’ hands.
Yet I believe many of you here have experienced similar pitfalls—broken demand transmission, uncontrolled project timelines, chaotic cross-departmental collaboration. These issues aren’t unique to any single OEM; they’re shared pain points across the automotive industry.
If in the past we managed isolated tasks, now we must connect them into a network, managing end-to-end value flows.
But understand this: A whole-vehicle project has never been a linear process “from A to B.” It is a three-dimensional network spanning R&D, production, quality inspection, and supply chain.
Therefore, what intelligent automotive companies need isn't merely a form-filling tool, but an integrated platform capable of intelligent evolution—one that scales up and down, reaches in and out: upward, providing strategic oversight; downward, enabling detailed execution; outward, capturing external feedback; inward, penetrating specific delivery nodes. Only such an engine can truly let processes speak and data drive decisions, turning a complex web of collaboration into a unified force.
It resembles a powerful, sophisticated “operating system.” On our journey to find this “operating system,” we were extremely cautious—until July last year, when we discovered Feishu Projects and deeply integrated it with our adopted IPD (Integrated Product Development) philosophy.
To us, these weren’t just features—they represented the potential to standardize innovation. Thus, we decided to introduce Feishu Projects at Avatr.
Looking back, that decision was pivotal. Over the past year-plus, Avatr hasn’t just been a user—we’ve been firsthand witnesses and participants in the rapid growth of Feishu Projects’ capabilities within the IPD domain. Therefore, I feel honored to represent Avatr and share our genuine experiences and valuable insights from the past year, focusing on three core dimensions: stability, intelligence, and safety of whole vehicles.
End-to-End Management, Dynamic Linkage, Drill-Down Capability
I believe you all understand: Whole-vehicle development is like a team marathon lasting one to two years—long in cycle and highly collaborative.
To fundamentally enhance project stability in vehicle development, Avatr made a key innovation in our R&D management system—establishing a clear and efficient “multi-level project planning system” to enable holistic management, real-time linkage, and drill-down visibility.
We’ve built a strategic-level planning system on Feishu Projects, achieving full oversight and focus on core objectives. Using swimlane diagrams and planning tables, we set “key milestones,” clearly defining progress targets, core deliverables, and key indicators at each node, along with strict review mechanisms to ensure alignment.
In addition, we leverage new capabilities such as task decomposition and hierarchical structuring to build Level 1 and Level 2 plans:
Level 1 Plan—Departmental collaboration, goal breakdown: Decompose high-level goals into specialized areas such as design and engineering, clarifying core tasks and upstream/downstream commitments, ensuring departmental goals align with overall vehicle objectives.
Level 2 Plan—Precise implementation, accountability assigned: Further refine the Level 1 plan down to secondary departments or core teams, specifying detailed task assignments, timelines, and responsible individuals, making execution transparent and responsibilities clear at every stage.
Within Feishu Projects, these three levels of planning are no longer separate data tables—they are alive, dynamically linked, and fully drillable. From the strategic objective of “Start of Production,” I can drill down directly to the delivery status of a specific component.
Our plans are complex, implying numerous deliverables requiring strict control. How do we ensure nothing is missed, while maintaining standardization and centralized management?
Now, we’ve transformed past lessons learned and accumulated best practices—the “mines” we once stepped on—into a “Deliverables Template Library” within Feishu Projects.
When launching a new project, nearly 1,500 key deliverables—including checklists, templates, and responsible parties—are automatically generated and automatically linked to underlying plan nodes, enabling process monitoring and predictable outcomes.
Furthermore, we’ve introduced AI-powered “pre-inspection” of deliverables: Before engineers submit their work, AI alerts them if content is incomplete or key information is missing. We’ve also implemented mandatory review workflows—only after completion can a plan node advance to the next phase.
Beyond that, this “template library” continuously evolves, incorporating new R&D insights and best practices into updated standards.
What has this achieved? We’ve converted our most valuable “tacit knowledge” into actionable “explicit assets,” improving the efficiency of our deliverable review and sign-off process by 30%.
This is not just time saved—it turns “past experience” into “confidence for future innovation.”
Once plans and deliverables go online, we gain something invaluable: real-time, effective data. Based on this, we’ve established a rigorous metrics and goal management system, allowing data to guide decisions instead of relying on “gut feelings” or “lagging reports.”
For example, we use the “OK to Buy” metric, setting over 1,000 checkpoints throughout key stages of vehicle development. Only when all relevant technical domains meet criteria can the process proceed. Feishu Projects helps us institutionalize this principle while supporting conditional approvals for exceptional cases.
We also have a goal management system—for core milestones such as SOP (Start of Production)—where key targets are tightly bound to the vehicle launch blueprint. Progress is visualized in real time, deviations trigger automatic alerts, ensuring every major decision stays precisely aligned with the original roadmap throughout the entire development lifecycle.
Running an “Automated Relay” Through Processes
The core of intelligent vehicles is “software-defined cars,” posing significant challenges for OEMs. Hardware development takes years, while software iterates weekly—leading to mismatches and redundant efforts. For instance, if the same autonomous driving algorithm is used across five vehicle models, do we really need to break down requirements and run processes five separate times?
Avatr’s solution begins with “turning technical assets into reusable strategic capital.” We’ve built a “R&D Technical Asset Platform” using Feishu Projects. Through a hierarchical functional tree structure, we achieve comprehensive control over functions and requirements across all projects. Any change in the function library propagates instantly to all relevant projects. Crucially, functional requirements can be one-click linked to multiple vehicle models, automatically entering the R&D pipeline. We also have a public test case plaza where new models can directly reuse mature technologies like autonomous driving perception algorithms.
Today, PXD analysis time per person has decreased by 90%, function decomposition time by 70%, and project function list alignment time by 50%. This platform-based control replaces manual, labor-intensive collaboration with intelligent, automated linkage.
But to truly make software processes “run,” we also need a “digitally automated middleware.” Using Feishu Projects’ workflow engine, we’ve launched an “automated relay.”
On one hand, we’ve moved the entire process—from Voice of Customer (VOC) to product delivery—onto Feishu Projects, vertically integrating the full chain.
But the key isn’t just “moving online”—it’s automation. Leveraging Feishu Projects’ workflow engine, we’ve embedded countless “automation triggers” across hundreds of thousands of R&D data points.
For example, upon receiving VOC input, AI first pre-tags and categorizes the requirement; once a “user story” passes review, the system automatically creates corresponding development and testing tasks in relevant R&D and testing projects; after test validation completes, the status is automatically fed back, pushing upstream requirements into a “closed” state. This entire chain elevates resource allocation from “experience-based decisions” to “data-driven decisions.”
On the other hand, we integrate over ten specialized systems—including development tools, design tools, and AI tools—via the open platform, achieving horizontal alignment.
Supplier drawings can be directly linked to corresponding R&D tasks in Feishu Projects; production line quality issues can trigger work orders with one click, assigned directly to the responsible design engineer.
In essence, we’ve handed off massive volumes of requirement transfers and status synchronization tasks to automated workflows, realizing the principle: “Work lives across all systems, but people stay in Feishu Projects.”
Freeing Human Talent to Tackle Tougher Challenges
“A 1% quality issue could lead to a 100% brand crisis.” Quality is the hardest standard at Avatr—and across the automotive industry.
Therefore, we’ve built a three-layered, full-lifecycle quality management defense system in Feishu Projects, shifting from “reactive firefighting” to “proactive immunity.”
The first line of defense is prevention. We’ve transformed historical quality issues into a “Lessons Learned Repository” within Feishu Projects. When a new project reaches certain phases, the system automatically pushes relevant risk alerts.
The second line of defense is in-process interception, achieved through quality gate milestones embedded in Feishu Projects’ workflows.
The third line of defense is closed-loop resolution. When non-conformities arise, Feishu Projects creates high-priority “issue tickets,” links them to root cause analysis templates, and pushes them to responsible engineers’ to-do lists. Accompanying dashboards enable dynamic tracking of resolution progress. Issues that previously took 14 days to resolve are now required to close within 72 hours, boosting overall management efficiency by 50%.
We’ve also equipped ourselves with an “AI teammate”—an AI-powered virtual quality engineer, internally nicknamed the “tireless night watchman.” As soon as an issue is logged, AI proposes solutions based on the TRIZ contradiction matrix; during the improvement phase, the virtual engineer uses the 8D model to verify corrective actions in real time. Compared to traditional manual methods, AI checks exceed 90% effectiveness, freeing up human resources to tackle more complex technical challenges.
Project Management Resilience Is Built With Feishu Projects
From July last year until today, colleagues at Avatr have been pleasantly surprised: Feishu Projects is far more than a simple project management tool—it is a powerful digital collaboration platform.
To date, Avatr has gone live on Feishu Projects with key scenarios including whole-vehicle program management, multi-level software development workflows, and a three-tiered quality control system. We are currently migrating additional vehicle lifecycle scenarios such as GTM (Go-To-Market), hardware-software co-development, and quality blueprints onto this platform.
We also plan to expand into supply chain optimization and AI-driven decision-making scenarios, continuously advancing Avatr’s digital transformation.
These scenarios are not isolated “dots,” but interconnected via Feishu Projects into a seamless “end-to-end chain—from customer demand to vehicle delivery.” For example, if users request “a smoother intelligent cockpit,” we can trace the path directly from market insight to R&D to production, with every step visible and trackable.
Today, 90% of Avatr’s core innovation projects operate smoothly on this R&D management hub.
In fact, in this VUCA era—volatile, uncertain, complex, and ambiguous—I often reflect: What truly constitutes an automotive company’s “moat”?
I believe the greatest challenge in building a good smart car isn’t a single technological breakthrough, but rather strengthening the organization’s overall capability—the synergy of efficient organization, processes, and mechanisms. To achieve this, we need a unified platform where “software agility, project stability, and collaboration efficiency” can operate organically together.
Just as a car drives steadily only when its engine, transmission, and chassis work in harmony, in whole-vehicle development, the “speed” of software, the “stability” of projects, and the “smoothness” of collaboration must be synchronized—requiring a central driving force. And Feishu Projects is precisely Avatr’s “super process engine.”
Whether in the automotive industry or beyond, sustainable success depends on this kind of project management resilience—the ability to “quickly adjust, precisely execute, and continuously optimize” regardless of external changes. Avatr’s practice shows this resilience isn’t “declared,” but “built” together with Feishu Projects.
The future of intelligent vehicles depends on collective industry effort. By sharing our Feishu Projects experience, we’re not claiming it’s the “only answer,” but rather illustrating this point: Only when more enterprises strengthen “collaboration efficiency, project resilience, and intelligent knowledge accumulation” can China’s intelligent vehicle industry truly upgrade—not by “building more cars,” but by “building better ones.”
The race for intelligent vehicles is long. What matters isn’t “who runs fastest,” but “who runs longest.” Avatr’s practice proves that in an uncertain era, “stable project management capability” is the strongest foundation. Only then can intelligent vehicles deliver more stable, smarter, and safer mobility experiences to users.
That concludes my sharing. Thank you all!