AI / LLMChina
← Sticker Wall
JOURNEY · DETAIL
Wave Speed AI - Wawa Writing: PMF Breakthrough of Vertical AI Writing Tools
Founded: Jiang Yuchen · Beijing Wave Speed Intelligent Technology Co., Ltd.
JOURNEY
Key Fields
FIELD STAMPSIndustryAI / LLM
RegionChina
ScaleSME
ChannelOther
Origin
Founder Jiang Yuchen previously led NLP-related R&D at Sogou. In 2022, observing that general AI writing tools commonly failed to align with domestic user habits and lacked competence in vertical scenarios—particularly in niche domains like workplace government documents and marketing copywriting where mature products were absent—she assembled a team to found Wave Speed AI, launching Wawa Writing to focus on vertical office writing scenarios.
Milestones
2022
Inception Stage Turning Point
Completed MVP development for Wawa Writing version 1.0, with core features covering niche writing scenarios such as student essays, workplace official documents, and reflective summaries. Achieved 23,000 registered users in its first month online with a paid conversion rate of 12.3%, validating real user demand for vertical AI writing scenarios and confirming the initial product development direction.
2023
Growth Exploration Stage Failure
The team attempted to expand into the online fiction writing track, investing 220,000 RMB to develop online literature-specific style templates and outline generation features. After launch, it gained only 12,000 new registered users with a paid conversion rate of less than 0.8%. Core users reported that the features did not meet actual online fiction creation needs, and after 3 months of operation, the team decided to cut this feature line to avoid wasting resources.
2023
Growth Stage Turning Point
Secured tens of millions of RMB in Angel round financing led by IDG Capital, with funds primarily allocated to expanding the official document corpus for government-affiliated units and developing compliance verification features. Paid users grew by 210% month-over-month that month, annual recurring revenue (ARR) surpassed 12 million RMB, and team size expanded from 12 to 35 members.
2024
Expansion Stage PMF
Launched enterprise-oriented marketing copywriting SaaS services catering to merchant marketing needs on platforms like Douyin, Meituan, and JD.com, featuring built-in promotional copy, product detail page, and short-video script templates. Signed 87 enterprise clients in the first month online, with an annualized contract value of 7.8 million RMB. Average daily user usage duration increased from 3.2 hours to 4.7 hours, and paid retention rate reached 72%.
2025
Adjustment Stage Transition
General foundation models such as Baidu Ernie Bot and Alibaba Tongyi Qianwen launched free official document writing features, causing Wawa Writing's paid user churn rate to hit 18% that month, with revenue declining 22% month-over-month. The team urgently adjusted its strategy, launching 'AI generation + manual compliance review' value-added services while expanding procurement channels targeting local public institutions and state-owned enterprises. Recovered 72% of churned users within 6 months, and full-year ARR for 2025 exceeded 120 million RMB.
Turning Points
- Cut the generalized online fiction writing track in 2023 and focused on vertical office writing scenarios, avoiding direct competition with general foundation models on general capabilities.
- Shifted from consumer-facing personal tools to enterprise-facing B2B SaaS services in 2024, finding the second growth curve to break through growth bottlenecks.
- Introduced manual value-added services in 2025 to counter the impact of general foundation models, securing the core paid user baseline.
Failures & Pitfalls
- Blindly expanded into the online fiction writing track in 2023; after investing 220,000 RMB in R&D resources, user feedback was extremely poor, leading to the cancellation of the feature line within 3 months.
- Tried launching consumer-facing entertainment features like AI poetry and AI love letters in early 2024, which failed to match core user needs. Monthly active users grew by only 0.8% after launch, and all features were removed within 2 months.
- Failed to timely anticipate the feature iteration pace of general foundation models in early 2025, resulting in massive paid user churn and a 22% month-over-month drop in revenue.
关键成功要素
- The core moat of vertical AI tools lies in scenario know-how rather than general model capabilities; one must deeply cultivate proprietary capabilities in niche domains.
- Deploy compliance corpora and content verification features ahead of time to mitigate policy risks associated with AI-generated content.
- Bind with B2B enterprise customer needs to enhance user paid retention and average revenue per user (ARPU).
- Flexibly adjust product strategies, avoiding direct competition with general foundation models on general capabilities by pursuing a differentiated route.
Lessons
- PMF validation for vertical AI products should first focus on core scenarios and expand boundaries only after validation, avoiding generalized products from the start.
- Product feature iteration must revolve around the real needs of the core user group, avoiding blindly following trends to expand into unrelated popular tracks.
- Faced with the dimensional strike of general foundation models, swiftly identify differentiated value points, such as human-in-the-loop services that general models struggle to cover.
- Financing pacing must match business development needs, retaining sufficient cash flow to weather industry fluctuations while avoiding over-financing that dilutes equity.
Core Data
- 2025 Annual Recurring Revenue (ARR)：120 million RMB (publicly disclosed figures, independent verification unconfirmed)
- 2025 Paid Users Count：320,000 (publicly disclosed figures, independent verification unconfirmed)
- Average Daily User Usage Duration：4.7 hours (publicly disclosed figures)
- Angel Round Financing Amount：Tens of millions of RMB (publicly disclosed figures, independent verification unconfirmed)
- 2025 Enterprise Clients Count：1,200 (publicly disclosed figures, independent verification unconfirmed)
- 2025 Paid Retention Rate：68% (publicly disclosed figures, independent verification unconfirmed)
Competitors / Peers
Domestic AI writing tools in the same track include Metaso, Gaoding AI Writing, and Tencent Zhiying. Among them, Metaso focuses on academic writing scenarios, while Gaoding AI leans more toward bundled services for marketing material design and writing. Wawa Writing's advantage lies in its deep cultivation of vertical scenarios for workplace official documents and enterprise marketing copy, featuring an exclusive corpus and compliance verification capabilities better tailored to domestic user habits, with a paid conversion rate in vertical scenarios approximately 40% higher than general writing tools.
Sources
- https://m.163.com/dy/article/J8K47MR905566TJ2.html
- https://news.qq.com/rain/a/20240803A028B900
- https://m.36kr.com/p/2193840525739905
- https://www.36kr.com/p/3848599219885313
- https://www.36kr.com/p/3807109385641993
- https://m.163.com/dy/article/J8K47MR905566TJ2.html?spss=adap_pc
- https://36kr.com/p/3778121523025154?f=rss
- https://news.qq.com/rain/a/20260523A0758U00
- https://36kr.com/p/3848599219885313
- https://txtmix.com/posts/video/yangmeng-anker-consumer-electronics-ai-2026/
- https://news.qq.com/rain/a/20250923A01CQY00
- https://36kr.com/p/3886914244360712
- https://36kr.com/p/3810099235053318
- https://m.163.com/dy/article/KV58AJAB05119C3G.html
- https://news.changsha.cn/cslb/html/111874/20250409/197856.shtml
- https://www.toutiao.com/article/7640873939924779547/
- https://www.toutiao.com/article/7483070454056370728/