# The New Consensus in Autonomous Driving: Lean AI Production with Feishu
## Industry Outlook for 2025
By **2025**, the autonomous driving sector is poised for **rapid growth**:
- **L2 Assisted Driving** penetration will surge
- Momenta’s urban NOA will take a leading market share
- Horizon Robotics’ “Journey” chips expected to exceed **10 million units** shipped
- Deeproute.ai’s solutions deployed in **130,000+ vehicles**
- **XPeng** and **Li Auto** will accelerate toward **L4 autonomy**
- **Robotaxi Expansion**:
- Pony.ai: aiming for a **1,000-vehicle** fleet
- WeRide: autonomous driving licenses in **7 countries**
- Neolix: delivered **10,000+** unmanned delivery vehicles
Despite diverse backgrounds and approaches, these **“physical AI” pioneers** share a new common ground:
**They are leveraging Feishu (Lark) for AI-driven lean production.**
---
## Why Lean Production for AI?
**Lean production**—born in the automotive sector—is about removing waste, reducing costs, and iterating faster.
In the **AI era**:
- **AI becomes both the tool and the product**
- Applying lean principles to AI enables:
1. **Rapid, up-to-date knowledge acquisition**
2. **Automated and standardized R&D**
3. **Minimized waste of human effort and intellectual assets**
### Leading Adopters
- **Horizon Robotics**
- **Momenta**
- **NavInfo**
---
## Case Study 1: Horizon Robotics — Lean Knowledge Management
**Challenge:**
Horizon generates **700,000+ documents annually**. A single project (HSD urban driving) creates ~**80,000 documents**.
**Solution:**
- Built knowledge management entirely on **Feishu**
- Created project-specific knowledge bases for **continuous iteration**
- Integrated **AI-powered knowledge Q&A**
- Enabled **permission management + security levels** using AI
- Developed **low-code document management platform** via Feishu aPaaS:
- Connected documents to actionable tasks
- Achieved **80%+ closure rate** in R&D tasks
**Impact:**
- Faster access to critical insights
- Secure sharing
- Increased execution speed
- Ecosystem adoption ("If Horizon uses it, partners use it too")
---
## Case Study 2: Momenta — Self-Driving R&D Process
**Focus:** Process automation with Feishu project management
**Highlights:**
- 1-click bulk import of development requirements
- Reduced time from **half a day to 10 minutes**
- Road-testing incidents uploaded instantly
- Automatic issue list creation in Feishu Projects
- AI clusters issues by scenario and pre-analyzes causes
---
## NavInfo & The AI Efficiency Pioneer Competition
**Initiative:**
- Internal competition showcasing efficiency improvements
- Co-hosted with Feishu to spread best practices
**Example: MapAuto's AI Group Message Summarization Agent**
- Built with Feishu low-code aPaaS + aily intelligent agent
- Converts fragmented chats into structured knowledge
- Reduces repeated problem-solving time
**Example: Map Navigation Business Center's AI Patent Agent**
- Multimodal interaction for patent disclosure drafting
- Conversational Q&A for application field and benefits
- AI extracts novelty points and keywords from research
- Full draft generated in Feishu docs for collaborative edits
- Supports AI meeting notes and innovation extraction from images
**ROI:**
Saved **320 hours/year** across 100 patents
Reduced R&D costs by **300,000+ RMB**
---
## Efficiency Imperative in Autonomous Driving
### Unique Constraints
- Need for **rapid iteration** without compromising safety
- High **cross-department coordination complexity**
- Traditional tools (calls, emails, dispersed chats) create silos and waste
**Example:**
WeRide replaced spreadsheet-based repair tracking with **Feishu Multi-dimensional Tables**:
- Expanded from **2 to 57** data points
- AI auto-generates monthly summaries & issue detection
- **80% fewer** repair issues in H2 compared to H1
---
## Market Pressure: L2 vs. L4
- **L2 Penetration** (Jan–Jul): 63% in domestic passenger cars
- Expected **100% L2 rate by 2030**
- **Robotaxi “first year of mass production” in 2025**
- Tesla's entry intensifies capital inflow and competitive speed
---
## Lean AI Production: The Strategic Shift
Feishu's **lean AI toolchain**:
- Minimizes wasted knowledge
- Activates assets faster
- Accelerates R&D cycles
**Supply Chain Effect:**
- Involvement from core players (Horizon Robotics, Pony.ai)
- Secure external collaboration with permission control
**Beyond Autonomous Driving:**
- Adoption among embodied intelligence leaders like Zhiyuan Robotics, UBTECH, Galaxy Universal, Songyan Power
---
## Complementary Ecosystems: AiToEarn
[AiToEarn官网](https://aitoearn.ai) — **Open-source global AI content monetization platform**:
- AI content generation, cross-platform publishing, analytics, and ranking
- Supports platforms: Douyin, Kwai, WeChat, Bilibili, Xiaohongshu, Facebook, Instagram, LinkedIn, Threads, YouTube, Pinterest, X (Twitter)
- [AI模型排名](https://rank.aitoearn.ai): benchmark models
**Parallel Impact:**
- Feishu boosts **enterprise R&D efficiency**
- AiToEarn boosts **creator monetization efficiency**
- Together, they show how **lean AI workflows** drive both *innovation* and *market presence*
---
## Key Takeaways
- **Lean AI production** is becoming the default across autonomous driving
- Feishu is at the core of this transformation:
- Manages knowledge at scale
- Automates workflows
- Enhances collaboration securely
- Early adoption of **advanced productivity tools** confers competitive advantage
- As Jensen Huang of NVIDIA says:
> *The next wave of AI will be physical AI.*
Those who embrace lean, AI-powered tools now will lead in that future.
---
