At the 2025 Feishu Project Product Launch, Fan Xiaojun, Senior Director of R&D at Momenta, shared practical insights and explorations into building an efficiency engine for intelligent driving algorithm development. He revealed core strategies that enabled Momenta to support a tenfold increase in vehicle models without significant team expansion.
Momenta has established a fully traceable workflow via Feishu Project, covering vehicle-end testing, issue management, requirement management, and acceptance management—achieving standardization, process optimization, and automated closure. Together with Feishu Project and Gaoyuan Huaxin, they co-developed an ASPICE plugin compliant with international OEM certification standards. The high flexibility of Feishu Project empowered Momenta to evolve from “importing SPA” to achieving the ideal state of “standardized input and output (SPA),” making each iteration of the flywheel large model more robust and driving intelligent driving R&D to become more efficient, stable, and precise.
Better AI, Better Life
Founded in 2016, Momenta is a globally leading intelligent driving technology company with the mission of "Better AI, Better Life." It strives to create a better life through breakthrough AI technologies, realizing scalable autonomous driving and making transportation safer, more convenient, and efficient.
The company has proposed a "Three-Decade Vision": saving one million lives in ten years, liberating 100% of human driving time in another decade, and doubling logistics and mobility efficiency within twenty years. Momenta continuously delivers technological solutions for safety assurance, free mobility, and doubled efficiency. Its technologies have been applied in vehicle models from major domestic and international brands such as SAIC, GM, and Mercedes-Benz, making it a preferred partner for many automakers.
Built on its "end-to-end flywheel large model," Momenta is creating stronger and more reliable intelligent driving solutions for global partners, accelerating the global deployment of smart mobility.
A Flywheel with Two Legs Strategy
Momenta implements a "one flywheel, two legs" product strategy. The first leg is intelligent assisted driving, which accumulates massive multi-scenario data through user usage. The second leg is scalable autonomous driving Robotaxi, aiming to launch the industry's first pre-installed mass-produced Robotaxi, significantly reducing per-unit costs.
The "one flywheel" refers to the flywheel large model, which operates at high speed supported by massive data, a closed-loop automated toolchain, and data-driven algorithms.
The Flywheel Large Model 5.0, launched in 2023, was the industry’s first single-stage end-to-end mass-production large model, introducing a human-like memory mechanism combining "long-term memory + short-term memory." The upcoming R6 model, set for mass production in 2024, will be China's first end-to-end large model based on reinforcement learning and truly deployed in real-world applications. It can autonomously explore driving behaviors in simulation environments, learn from successes and failures, and potentially surpass human drivers in safety performance.
Since its founding in 2016, Momenta has achieved rapid growth in intelligent driving deployments and partnered vehicle models. The number of mass-production collaboration models increased from just one in 2022 to over 130 by April 2024, with software installed on more than 400,000 vehicles, continuing to grow rapidly.
Automate Any Task Repeated Two or Three Times
Facing the challenge of surging customer projects and vehicle models while constrained R&D team growth, Momenta’s secret lies in advanced software architecture, leading algorithms, strong engineering capabilities, and highly efficient software development processes.
The company advocates the “Input-to-Output SPA” concept: S (Standard) represents standardization, P (Process) stands for process optimization, and A (Automation) means automation. This system ensures accurate requirement input without omissions, high-quality version outputs, and all processes running under a unified framework capable of quickly adapting to customer needs.
Momenta has a rule: any task repeated two or three times must be automated using tools. Although self-developed systems were considered, the team ultimately chose Feishu Project because its platform capabilities closely matched Momenta’s requirements for R&D process tools.
Feishu Project offers two outstanding advantages: first, flexible and efficient workflow configuration, supporting low-code/no-code rapid setup and adjustment of hundreds of complex workflows, significantly lowering digitalization barriers; second, platform openness and secondary development capabilities, enabling seamless integration across all R&D stages via standard interfaces for smooth information flow—outperforming foreign tools in local scenario adaptability.
After a two-week trial, Momenta decided to fully adopt Feishu Project—a true “mutual fit,” where Momenta had clear needs and Feishu Project delivered the right capabilities.
In addition, Momenta has developed multiple automation plugins to enhance internal R&D efficiency, achieving true “autonomous driving” in R&D workflows.
Connecting Six Core R&D Processes
1. Traceable Workflow: Native bidirectional traceability in Feishu Project transforms high-risk, high-complexity processes—from customer requirements to delivery—into transparent, controllable, and verifiable chains. This not only meets ASPICE requirements but also forms the foundation for safe intelligent driving deployment. After implementation, communication frequency between product managers and clients dropped from 20 times per week to 1–2 times, with information automatically flowing through the system.
2. Road Test Management System: Built on Feishu Project, this system covers the entire process from test planning and case design to execution. Test tasks are synchronized to in-vehicle displays, allowing safety operators to follow instructions and report issues with one click. The system automatically collects vehicle data, generates issue lists, and produces complete reports. One person can now supervise up to 10 vehicles, increasing efficiency tenfold and enabling limited resources to cover infinite real-world scenarios.
3. Intelligent Issue Management: Leveraging AI technology, the system enables full-cycle automated closed-loop issue management. Once issues are reported from the vehicle or customers with one click, the system automatically clusters them (e.g., consolidating 4,000 reports into over 300 key items), reducing R&D workload. It performs AI pre-analysis and outputs possible causes in图文 form for reference, supports one-click feedback to optimize models, and greatly improves processing efficiency.
4. Efficient Requirement Management: Solves the problem of diverse customer requirement formats and inefficient verification. The system supports one-click batch import of various formats, reducing the time needed to process thousands of requirements from half a day to under 10 minutes. Product managers digitally map customer requirements to mainline requirements, allowing R&D teams to focus on core progress with more efficient synchronization and fewer meetings.
5. Multi-Layer Release Gate Mechanism: Establishes strict release gates as safety red lines and the cornerstone of mass production. Every code commit requires code review and unit testing; modules undergo bench validation; multi-module integration is verified through simulation and real-vehicle tests; final mainline versions undergo additional audits before delivery to customer projects, ensuring every released version is reliable and compliant.
6. ASPICE Plugin Certification: Co-created with Feishu Project and Gaoyuan Huaxin, this ASPICE plugin meets international OEM supply chain entry standards. It supports one-click import of original customer requirements in Reqif or Excel format, establishes document version and baseline repositories covering the entire R&D process, and automatically generates audit-compliant documentation evidence.
Turning Complexity into Simplicity, Ambiguity into Clarity
The core value of Feishu Project lies in solidifying the foundation of R&D processes—standardization, process optimization, and automation—enabling the ideal state of “from input to output.” As an efficiency engine, it makes each iteration of the flywheel large model more robust, and R&D work faster, more stable, and more accurate.
Although not perfectly aligned with every expectation (e.g., some metric views require custom plugin extensions), the openness and flexibility of Feishu Project are deepening its collaboration with Momenta, simplifying complex tasks, clarifying ambiguous scenarios, and visualizing invisible work—unlocking infinite possibilities.
Currently, Feishu Project has integrated over 20 internal systems, developed more than 50 plugins, and sees over 1.2 million daily API calls, enabling massive automated data flow and greatly reducing labor costs.
Internal surveys show significant efficiency improvements across all stages. More importantly, the value of digital tools goes beyond mere efficiency. By freeing engineers from repetitive tasks, they can now focus on next-generation intelligent driving model experiments, safer algorithm research, and pushing the boundaries of artificial intelligence. The energy once trapped in meetings and spreadsheets is now reshaping the future of the industry.
This is the true essence of building an “efficiency engine”—freeing human potential to focus on higher-value innovation and thinking.
As CEO Cao Xudong stated, we are now at the cusp of the “Intelligent Driving Moore’s Law,” where rapid large-model iterations are reshaping the industry. Momenta is committed to going beyond Moore’s Law and advancing democratization in intelligent driving.
To achieve industry transformation, it’s not enough to break technological limits—we also need an “equality revolution” driven by ecosystem collaboration, relying on all participants across the industry—from engineers to data labelers, from automakers to suppliers—to jointly improve efficiency and define the pace of technological adoption.
Therefore, we need tools like Feishu Project, as well as collective efforts from the entire industry.
Returning to our初心: Better AI, Better Life. Through breakthrough AI technologies, achieving scalable autonomous driving so everyone can enjoy safe, convenient, and efficient mobility—that is the fundamental driving force behind Momenta’s continuous advancement.
Click here to access highlights from the 2025 Feishu Project Product Launch!