When a new car hits the road, it's no longer just sleek lines or raw power that win people over. From confident braking to effortless navigation, intelligent driving is reshaping the relationship between humans and their cars—turning machines into smart partners that understand and adapt to their drivers.
ZHUOYU is leading this transformation with a rare strength: end-to-end, in-house R&D spanning both hardware and software. This unique advantage fuels the company's ability to keep innovating and delivering at speed. Yet what truly powers ZHUOYU's progress is not just technical capability, but the deep integration of Meegle into its R&D management.
In early 2025, ZHUOYU embedded Meegle into its core development process. By introducing intelligent milestone tracking and a connected data asset chain, the company cut its model development cycle by 32% and boosted issue resolution efficiency by 52%. These gains provided the critical foundation for ZHUOYU to pass Volkswagen's ASPICE CL2 Software Assessment—making it the first domestic supplier to earn this achievement.
The Tool Dilemma: Fast Iterations vs. Rigid Processes
Developing an advanced driver-assistance system is far more complex than it appears. A complete system requires tight integration across interaction design, software development, AI model training, embedded systems, and hardware architecture—where a single change can ripple across the entire platform.
One of ZHUOYU's key breakthroughs was deploying an explainable end-to-end model in mass-produced vehicles, including the Hongqi Tiangong series and Volkswagen models. By connecting perception, prediction, decision-making, and planning modules, this model minimized information loss, elevated system performance, and, through interpretable intermediate outputs, allowed users to better understand and trust the vehicle’s autonomous behavior.
Yet technical success also highlighted a management challenge. Intelligent driving projects are inherently agile, with rapidly evolving requirements, shifting dependencies, and frequent cross-team deliveries. Traditional tools quickly reached their limits.
Early attempts with Jira, for example, ran into fragmentation issues. "Jira is isolated from our communication tools, its workflow states are too basic, and it can't handle complex task breakdowns", admitted ZHUOYU's PMO. Development modules vary widely, and some—like model development—require over 20 distinct steps. Teams were forced to operate outside the system, leaving task progress and records out of sync. The question became: how could ZHUOYU digitalize and visualize, and deeply customize workflows while enabling large cross-functional teams to collaborate seamlessly?
The Turning Point: Rebuild the Digital Backbone of Innovation
ZHUOYU's approach to model development is fundamentally different from traditional software delivery. What makes this path distinctive is its closed-loop design—spanning everything from data collection to model deployment. Along the way, it brings together more than ten specialized roles across multiple stages, while ensuring strict process compliance and full traceability of every data asset.
"What impressed us most about Meegle was its flexibility, especially in workflow customization," said ZHUOYU's PMO. "Instead of dictating how we should work, it lets us design rich templates and connect nodes for different development paths. Independent teams can finally collaborate seamlessly in a single shared space. Requirements flow in, products iterate, tasks hand off—all without friction."
At the heart of this new way of working is a clear goal framework in tree view. Department-level strategies cascade into measurable objectives for each product line, which then break down into feature modules and finally into atomic user stories. For example, a high-level goal like "improve parking accuracy" becomes the sub-goal "optimize obstacle detection models" for the algorithm team, which in turn is executed through specific tasks such as "data collection" and "model training." This structured, transparent system keeps over 27,000 requirements moving smoothly, ensuring that strategy and execution stay tightly aligned.
250px|700px|reset
Once top-level goals were clarified, execution rhythm became the key. Facing ambiguous requirements and frequent cross-team handoffs, teams now manage work in two-week sprints, rigorously tracking dependencies and breaking down features into user stories. Meegle's customized visual workflows brought this rhythm to life.
Take embedded software development as an example: once a requirement is submitted, it automatically generates a feature, which decomposes into development and integration stories. While developers are still coding and self-testing, testing teams are already designing validation plans. At the review stage, the system automatically triggers testing tasks and sends notifications. Gone is the old "relay race" where teams waited anxiously for the baton. "Before, we had to wait for handoffs—it was always stop-and-go," recalled an R&D lead. "Now it feels like running side by side: every team sprints in its own lane, but we're perfectly aligned."
By transforming its tools from a rigid constraint into the backbone of innovation, ZHUOYU has bridged cutting-edge intelligent driving technology with equally agile management practices—allowing complex R&D to scale rapidly without sacrificing speed, precision, or cohesion.
250px|700px|reset
Unique Path: An End-to-End AI Model Development Chain
Model development for intelligent driving is nothing like traditional software delivery.
Take automated parking as an example. The process starts with data engineers collecting real-world driving scenarios. Algorithm researchers then train and optimize models based on that data. From there, models go through a strict pipeline: quantization to fit limited on-board computing power, deployment to hardware, large-scale simulation, real-road testing, and finally, a safety team's rigorous validation.
This pipeline isn't just complex—it's data-intensive and role-heavy. More than ten different functions are involved, from data engineers and algorithm specialists to testers and safety reviewers. Every stage follows a strict order, with dependencies and heavy documentation tying them together. And the technical constraints are unique: vehicles must run highly complex models in real time on limited hardware, without over-relying on cloud computing that could cause dangerous delays. These dual pressures have made innovations in quantization, deployment, and engineering optimization essential for success.
Previously, this precision-driven process was fragmented across multiple spreadsheets. Reports and data versions often got out of sync, wasting hours on manual clarification and slowing down delivery.
With Meegle, ZHUOYU built a fully digitized development chain. Key stages and roles are now connected in a visualized collaborative workflow, integrating what used to be scattered across documents, schedules, and acceptance checklists into one seamless system. Meanwhile, non-essential nodes are automatically skipped, keeping the standard path lean and efficient, while preserving the flexibility to handle more complex or exceptional cases.
For ZHUOYU's PMO team, the biggest breakthrough has been dependency management. Once a task is completed and accepted, the next stage is triggered automatically. Every acceptance report becomes a mandatory prerequisite before the workflow can advance—ensuring that no new task begins until the previous step has been rigorously validated. This creates a clear, auditable record of every release, with timestamps that expose bottlenecks and make process optimization far more systematic.
All critical assets—data versions, test results, validation reports—are now embedded directly in work items, forming a complete, traceable digital asset chain. The results are tangible: model delivery cycles have been cut by 32%, and six categories of AI models can now be developed in parallel with stability—a leap forward in both speed and scale.
250px|700px|reset
Quality Safeguard: A Data-Driven Closed-Loop Defect Mechanism
In intelligent driving development, millions of kilometers of road testing are indispensable—but true competitiveness lies in how effectively teams can extract insights from data, resolve defects, and continuously strengthen both system reliability and user experience.
ZHUOYU has developed a data-driven closed-loop defect mechanism that fuses data streams directly with the development workflow. From the moment an issue arises to its final verification, every step is automated and traceable, ensuring problems are located precisely and resolved efficiently.
During real-world road tests, each vehicle continuously collects operational logs and abnormal event data, with the option to upload incidents on demand. After testing, this data is automatically transferred to a unified platform, where AI algorithms analyze and surface critical problem scenarios. Through a sorting mechanism, these issues are instantly linked to Meegle, which generates standardized defect tickets. In this system, issues are no longer dependent on manual reporting—the data itself raises the flag.
Once created, defects are managed entirely within Meegle where engineers handle them in their daily workspace. Each ticket can be traced back to the relevant requirement or functional module, providing two-way visibility between issues and development context. For repetitive or similar problems, the platform supports clustering and batch handling, cutting redundant analysis and improving efficiency.
Transparency and accountability are built in. A visualized Kanban board tracks every defect in real time, giving stakeholders an immediate, intuitive view of progress and bottlenecks. Automated notifications keep management updated on key milestones, enabling a full cycle of "detection → response → resolution → review" without blind spots.
The loop extends all the way to validation. Once a defect is identified, the R&D team can schedule and assign the repair tasks directly in Meegle. When the fix is completed and deployed, the system triggers the next step automatically: Once a fix is deployed, Meegle connects directly with the automated testing platform to replay the original problem scenario in simulation. By comparing pre- and post-fix behavior, the system confirms whether the defect is truly resolved and updates the ticket automatically. The result is a seamless chain of "discovery → analysis → repair scheduling → automated verification".
The results speak for themselves. Average defect resolution time has dropped by 52%, regression misses have fallen sharply, and recurrence of major issues is now close to zero.
In the era of data-driven development, issues are not merely fixed after the fact—they are anticipated, tracked, and verified with precision. With this closed-loop mechanism, ZHUOYU has built a robust safeguard for the high-quality delivery of intelligent driving products.
250px|700px|reset
Closing
Transformation doesn't always arrive with fanfare. As ZHUOYU's engineer reflected, "I used to spend every day drained by endless notifications and chasing people for updates. Now, the tool has become a 'silent collaborator', quietly handling routine work."
This shift was made possible by consolidating documents, tasks, and notifications into Meegle. R&D teams no longer juggle multiple disconnected systems; instead, a streamlined dual-platform model—Meegle plus specialized development tools—supports all workflow needs. This digital process innovation harmonized perfectly with ZHUOYU's landmark achievement in August: becoming the first Chinese supplier to pass the joint ASPICE CL2 Software Assessment from Volkswagen China, FAW-Volkswagen, and SAIC Volkswagen. This achievement, widely regarded as the "golden key" of international automotive software standards, validated not only ZHUOYU’s technical expertise but also the maturity and scalability of its global process system.
ASPICE CL2 imposes strict requirements on performance management, task management, and schedule tracking—all of which were fully supported by Meegle. From requirement decomposition and progress monitoring to risk alerts and quality gates, the platform provided a structured, auditable workflow. This ensured compliance with the world-class standards of the audit while establishing a reusable global process framework that enables rapid, coordinated innovation across projects.
In what many call the "first year of democratized intelligent driving", ZHUOYU's transformation has done more than streamline operations—it has reshaped the culture of intelligent driving development. Every iteration is now closely aligned with real-world road safety needs, embedding reliability and performance into every product release. By connecting digital tools, rigorous processes, and engineering creativity, ZHUOYU is accelerating the adoption of intelligent driving technologies, helping the industry move toward a smarter, safer, and more human-centered future of mobility.
