Feishu no longer independent, but ByteDance didn't lose; the SaaS industry truly lost

On July 30, 2025, an internal email from ByteDance ended Feishu's decade-long independent operation. There was no lengthy farewell letter, no launch event, just one sentence: "No longer having independent sales and marketing teams." From this point on, Feishu was officially integrated into the Doubao ecosystem.

Looking at the data, this decision wasn't sudden. In 2023, Feishu held about 6% of the Chinese collaborative office market share, while DingTalk and WeCom held 42% and 35% respectively. With a team size exceeding 8,000 people and estimated annual revenue under 3 billion RMB, by ByteDance's internal efficiency standards, the math didn't add up. But what truly caused Feishu to "disappear" wasn't losses, but ByteDance's reprioritization of AI.

I. Feishu Didn't Die, It Became Doubao's "Organ"

Many interpret this as Feishu's failure. But the internal logic at ByteDance is: Feishu no longer needs to sell SaaS to external customers; instead, it serves as AI infrastructure, providing a "workflow engine" for Doubao.

Look at Doubao's current product form: AI chat, document collaboration, project management, meeting minutes. Over the past two years, the Feishu team has done a lot of underlying development for these capabilities. But the problem is, selling Feishu separately means customers pay for SaaS and also buy Doubao AI—two separate bills. ByteDance discovered that what the market really pays for is an integrated AI + collaboration experience, not standalone IM or documents.

So the core action in the email is: Feishu's product team merges into Doubao, while sales and marketing teams are dissolved or reassigned. This means ByteDance has abandoned using the Feishu brand to compete with DingTalk and WeCom for enterprise clients, shifting to using Doubao to capture the AI workstation market.

Feishu Old Architecture:
Standalone Product -> Sales Team -> Enterprise Clients -> Paid Subscription

Doubao New Architecture:
AI Capabilities (Doubao) + Collaboration Capabilities (Feishu Underlying) -> Unified Product -> Enterprise Clients -> Pay per AI Call Volume

Behind this change is a fundamental shift in the revenue model. From SaaS subscriptions of 2,000–5,000 RMB/person/year, to charging by tokens and API call counts. The unit price might be higher, but customers feel they get AI + collaboration solved in one go, without needing to maintain two suppliers.

II. The "Independent Narrative" of the SaaS Industry Has Completely Bankrupted

Honestly, Feishu's ten years were the most expensive experiment in China's SaaS industry. ByteDance spent hundreds of billions to verify a cruel fact: in China, it's very hard for standalone enterprise-level SaaS products to make money.

It's not that the product is bad. I think Feishu is better than DingTalk and WeCom in terms of interaction experience, document collaboration, and OKR alignment. But the issue is that Chinese enterprises have extremely low willingness to pay, and large enterprises prefer customization over standardized products. Feishu took extreme toC product thinking to do toB, resulting in: customers feeling it's "good but unnecessary," because the free version is sufficient, and the paid version is too expensive.

More critically, after AI arrived, the pricing logic for SaaS changed entirely. Previously, software was sold with three-year contracts, charged per headcount. Now, AI is sold based on results and usage, allowing customers to switch anytime. After DingTalk launched its "AI Assistant" in 2024, monthly active users rose by 30%, but the paid conversion rate remained low. Because customers aren't willing to pay extra for the incremental value brought by AI.

ByteDance realized one thing: rather than spending effort educating the market to pay for Feishu, it's better to use Feishu's capabilities as a free foundation for Doubao, and monetize via Doubao's AI capabilities. Feishu is no longer a product; it's a cost center.

[!note] Ironically, the Feishu team just launched an AI product called "Feishu Intelligent Partner" in 2024, benchmarking Microsoft Copilot. Now, this product has directly become a module of Doubao, losing even its name.

III. In the AI Era, There Are No "Independent" Software, Only "Embedded" Capabilities

Here's a more straightforward analogy: Feishu is to Doubao what Apple's "Computational Photography" is to the iPhone. You don't buy computational photography software separately, but you pay a premium for the iPhone's camera. What ByteDance aims to do is make Doubao the "iPhone" of enterprise office work, with Feishu being the "camera."

From an industry signal perspective, this isn't an isolated case. Microsoft Teams is no longer sold separately but bundled with Office 365 and Copilot. Salesforce, after acquiring Slack, turned it into the interaction interface for its AI platform Einstein. The era of standalone collaboration software is accelerating its end under the AI wave.

What does this mean for developers? If you're still building standalone SaaS products, especially in collaborative office, documents,

Original link: https://www.tmtpost.com/8085035.html