Feishu Projects is now more flexible, powerful, and intelligent!
On September 10, at the 2025 Feishu Projects Product Launch Event, Tao Hong, Head of Feishu Projects, unveiled significant platform upgrades, introducing a new generation of Feishu Projects that is more flexible, more powerful, and more intelligent.
At the event, Hong stated: "This enhanced version of Feishu Projects completely encapsulates complexity—whether in scenarios, systems, or data—to deliver an ever-simple and efficient user experience. This is precisely the ultimate pursuit of 'mastering complexity through simplicity.'"
Below is the full transcript of his speech.
Standing here today, I’m reminded of my first visit to Liepin over three years ago. In Liepin’s meeting room, I told their team: 'Choosing Feishu Projects isn’t just about adopting a standardized software—it’s about choosing a product that grows alongside your business.' It was a great honor that Liepin chose to believe in this vision, becoming Feishu Projects’ very first SaaS customer. That trust marked Feishu Projects’ official step into serving external clients, reinforcing our belief that 'products have life'—a life that evolves in sync with customer needs and resonates with industry development.
Today, this 'vitality' has taken on a richer meaning. Take Momenta, a global leader in intelligent driving, as an example. Their use of Feishu Projects is not only deep but also highly creative—their level of flexibility even amazed me, exceeding what our product team had imagined. Here, I’d especially like to thank colleagues from Momenta, Genki Forest, and Avatr who joined us today, sharing their valuable experiences using Feishu Projects. On Feishu Projects’ journey, they are not merely deep users, but indispensable 'co-creators.' I look forward to your insights later! It is precisely this deep engagement from customers that enables Feishu Projects to better meet real-world industry demands. Serving outstanding customers like these is the most direct manifestation of Feishu Projects’ vitality.
Over the past year, this 'vitality' has translated into tangible growth metrics: we’ve added over 3 million lines of code, rapidly responded to and implemented more than 800 product and technical requirements; our automation capabilities run approximately 1.5 million times per day, freeing countless users from repetitive tasks so they can focus on higher-value innovation; our open platform continues to break new ground, with API calls reaching 20 million daily—a 67% increase year-on-year—indicating that more and more customers and partners are building their own business ecosystems atop Feishu Projects. Behind these numbers lies vivid proof of Feishu Projects’ continuous iteration and evolution, guided by our customers every step of the way.
Product Architecture: A Three-Layer Design Enabling Platform Capabilities
To meet the diverse project scenario needs across countless industries, we understand that the software layer must be sufficiently abstract and open. Unlike many other project management tools, Feishu Projects was designed from day one as a platform product, with its core competitive advantage being a robust and continuously evolving cloud platform. For end users, this platform represents a baseline of capability and experience—any solution built upon this strong foundation inherits equally exceptional functionality and usability, giving users confidence in the product’s reliability.
Feishu Projects’ architecture is clearly divided into three layers. At the bottom is the infrastructure layer—the solid 'foundation' that determines system security and performance thresholds. Above it lies the core capabilities layer, which revolves around three main pillars: workflows, views, and metrics. Through modularized capabilities, it supports the free construction of business models. Together, these two layers form Feishu Projects’ Core Platform. Upon this Core Platform 'base,' we can then, at the topmost application layer, quickly build diverse product solutions and industry-specific offerings tailored to individual customer needs, transforming underlying complex capabilities into simple, intuitive project management experiences for users.
We use 'visualized workflows' to organize thousands of nodes into clear flow maps, enabling managers to instantly spot bottlenecks and executors to clearly understand their responsibilities; we use 'high configurability' so enterprises don’t have to adapt to software, but instead can build custom models on demand; and we use a 'platform-based foundation' to package complex capabilities into simple tools.
Using this approach, customers can focus solely on 'complex business innovation,' while Feishu Projects simplifies the 'complex management process' for them—this is 'mastering complexity through simplicity.' We keep complexity within the platform and return simplicity to the user.
Today’s product upgrade embodies a deeper implementation of this philosophy, leveraging more flexible, powerful, and intelligent platform capabilities to directly address the complexities of R&D management.
More Flexible: Digitizing Complex Scenarios with Tailored Customization
This year, Feishu Projects’ visual workflow engine underwent a comprehensive customization upgrade, significantly enhancing its visualization capabilities. The newly introduced node information preview feature allows project managers to customize what information appears on each node—whether numeric values, text, or enumerated options—all fully configurable. With this function, nodes can now display not only task completion progress but also a concise, one-sentence summary of the latest developments.
Node buttons now support system-level actions such as 'rollback' and 'reassign,' while also allowing users to define custom buttons and rules. The open plugin framework further enables the creation of bespoke business logic. For instance, with the DevOps plugin shown in the image, clicking 'Link MR' instantly redirects to the integrated platform.
The fully customizable workflow engine also delivers seamless mobile compatibility, significantly improving usability.
The new Teams feature integrates seamlessly with existing enterprise organizational structures, enabling联动 between project and organizational management, and supporting virtual team creation. Whether in metric views or personnel scheduling, quick filtering or grouping by team level is now possible. Hierarchical permission management offers finer control over objects and details, meeting the精细化 management needs of large organizations and enabling personalized field and information access rights for different users.
Metric charts have undergone a major upgrade, introducing professional chart types such as radar charts, waterfall charts, and stacked charts, covering diverse data analysis and visualization needs. The new chart system supports theme customization, enriching the visual presentation of individual charts and significantly improving information delivery efficiency.
Beyond flexible model configuration, this year we’ve launched a new feature—Feishu Projects Light Apps—which best exemplify the platform’s flexibility.
Longtime Feishu Projects users know that multiple functional homepage entries can be configured in the collaboration space’s left navigation. Previously, however, these could only be system-preset templates or plugin applications requiring full-code development by IT teams.
The newly launched Feishu Projects Light Apps empower regular users to build business application modules without writing any code, allowing each functional homepage to perfectly match team and scenario usage habits. As easy as building blocks, these apps can be created quickly without development skills—just as effortlessly as configuring fields, roles, and workflows.
Take the Iteration Management Light App as an example: a Scrum Master can use rich official components to quickly assemble a 'Light App.' Components can communicate and interact—when a user manually filters iteration types, other interface components update in real time. With the introduction of page variables, functional components can better consume contextual environment information. Supported by page variables and component interactivity, functional homepages are no longer static dashboards, but truly 'living' applications.
In the collaboration space’s functional area, project administrators can add various 'Feishu Projects Light Apps,' ensuring smooth, tailor-made deployment for every scenario. Featuring free layout design, a rich ecosystem of components, page-level variable context, efficient component interaction, and no-code rapid development, Feishu Projects Light Apps greatly expand the platform’s ability to support diverse use cases.
More Powerful: Hiding System Complexity, Handling Challenges with Ease
'Avatr' means 'incarnation.' At Avatr Technology, it carries deeper significance—aiming to create another you in a smart parallel world, emphasizing 'emotional intelligence' and becoming 'an intelligent avatar that understands you.'
Here are some real customer usage statistics: a gaming client configured over 1,000 fields for a single work item type; an automotive OEM broke down a single process into over 1,000 nodes; an internet software company set up over 400 automation rules within a single workspace; Ruijie Networks stores over 8 million instances.
Such massive data volumes pose severe challenges to system stability and performance. Imagine if all these limits were concentrated in one customer: a single work item with 1,000 fields, a process with 1,000 nodes, over 1,000 automation rules, and storage of over 10 million instances—this level of complexity is increasingly approaching real-world scenarios. As a professional project management product, we must rise to this challenge.
To achieve this, our R&D team launched 'Project 4K,' where '4K' refers to the four aforementioned engineering benchmarks. To meet these goals, we re-architected the underlying data storage, adopting a decoupled storage-and-compute architecture to enable dynamic, multidimensional data distribution. Data is distributed not only by tenant but also by business data type across different storage nodes, theoretically enabling 'infinite' scalability. On the retrieval side, we employ a distributed hybrid data indexing system combining row-based and column-based storage, supporting arbitrary data types and dynamic queries via a schemaless architecture.
On one hand, we’ve dramatically increased capacity limits, elevating support for scenario complexity to a new level; on the other, while maintaining high-complexity support, we ensure user experience remains smooth and responsive, with continuous optimizations. Over the past year, our R&D team has invested heavily in performance improvements.
As a result, compared to a year ago, key page load speeds in Feishu Projects have improved by over 130%; work item detail pages with over a thousand fields now open instantly; metric charts with millions of rows render smoothly; the workflow engine’s frontend framework now supports up to 5,000+ nodes; and over 100 interaction refinements have been completed throughout the year.
More Intelligent: Bringing Complex Data to Life, Delivering On-Demand Insights
In recent years, the pace of AI breakthroughs has exceeded expectations, reshaping work styles, organizational forms, and even corporate core competitiveness. Many colleagues and customers have asked me: 'Tao, how does Feishu Projects view AI?' Today, I’d like to share our perspective.
We remain cautious about AI completely disrupting project management software. On one hand, such transformation may take a decade or even longer—it cannot be rushed. On the other, as a product only three years old, we still have substantial foundational work to do; blind investment might lead to diminishing returns. At this stage, Feishu Projects’ value proposition in the AI era manifests in three key areas:
First, serving as an enabler of enterprise AI readiness. Feishu Projects accumulates the most valuable 'fuel'—structured data. Process nodes, task progress, human effort inputs—these are not just static numbers, but the enterprise’s 'intelligence genes,' ready for AI to consume, mine, and feed back into decision-making.
For example, gaming client Moonton integrates Feishu Projects’ version management with DeepSeek, extracting multi-dimensional data such as versions, requirements, and defects, allowing Agents to quickly generate summaries and reminders for release notes.
Internally at ByteDance, R&D teams have excellent practices: AI analyzes structured R&D data accumulated in Feishu Projects to automatically generate R&D efficiency reports, identifying correlations among project progress, resource allocation, and quality metrics, and delivering actionable insights and recommendations.
Today, to help AI better understand Feishu Projects data, we’re officially launching Feishu Projects MCP. MCP is a service built specifically for AI Agents—not just a wrapper around OpenAPIs, but a fully AI-friendly open server that makes it easier for AI to call, consume, and understand data.
Second, becoming an action system for the AI era. Feishu Projects must evolve beyond being just a 'system that records data'—it must become an 'intelligent engine that drives action.'
Our much-loved visual workflow engine is quietly transforming. In the past, workflows drove human owners: people received notifications, broke down tasks, executed them, and manually updated results back into the system. Now, workflows are turning the platform from a record-keeping system into an action-driving system: when a node is reached, it’s no longer a person who takes over, but an AI Agent that automatically understands the business context, performs actions, and synchronizes results back into Feishu Projects to complete recording and progression.
For example, in an upgraded AI-powered test management practice, before quality validation, a Feishu Projects workflow triggers a 'generate test cases' node. A test case Agent automatically generates a predefined set of test cases, then advances the process to the QA validation node—enabling relay-style collaboration between AI and humans.
During this process, the test case Agent also analyzes other data within Feishu Projects to automatically identify high-risk scenarios and low-risk test cases, helping QA engineers focus their efforts and significantly boosting efficiency.
With Feishu Projects, humans and AI collaborate synergistically—each playing distinct roles while mutually enhancing one another.
Finally, Feishu Projects natively introduces practical AI capability packs.
AI fields can convert unstructured data into structured information with a single click based on preset rules; they can also interpret or update field values through natural language conversations, drawing from multi-dimensional contextual data.
AI Personal Weekly Reports help employees generate structured summaries—including weekly progress, potential risks, and next steps—with one click, based on task statuses, document plans, meeting notes, and key group chat discussions in Feishu Projects. This eliminates the tedious need to gather data across platforms, using AI analytics to distill fragmented information into actionable insights.
AI Resource Analysis quickly retrieves team resource input data over specified periods, accurately identifies anomalies such as over-allocation or scheduling conflicts, and provides optimization suggestions—transforming organizational effectiveness measurement and improvement from complex data processing into intuitive, intelligent analysis.
Ladies and gentlemen, this new Feishu Projects—more flexible, more powerful, and more intelligent—completely encapsulates complexity in scenarios, systems, and data, delivering to you an ever-simple and efficient user experience. This is the ultimate pursuit of 'mastering complexity through simplicity.'
Finally, thank you all once again for being here and for your continued support! Thank you!
Click Read Original to access highlights from the 2025 Feishu Projects Product Launch!