The Search Box Is Losing Its Monopoly

China's search engines are losing shoppers to chatbots. Doubao, ByteDance's AI assistant, counted 382 million monthly active users in June 2026 — more than Tongyi Qianwen (167 million) and DeepSeek (130 million) combined 1. Increasingly, that traffic is not typing product names into Baidu; it is asking an AI assistant what to buy, and the assistant answers with a short, confident list of brands. Whoever shapes that list controls a slice of Chinese consumer spending that Baidu-era search-engine-optimization firms never touched.

That shift has produced a new business category almost overnight: generative engine optimization, or GEO — engineering a brand's footprint across the training data, retrieval indexes and real-time web sources that large language models draw on when they answer a shopping question. PureblueAI (清蓝), a Beijing startup barely two years old, has become the category's most closely watched operator, and its cap table tells the more interesting story: it has raised capital twice in six months, from two backers with opposite motives.

A PR Conglomerate's Defensive Bet

The first check came in September 2025, when BlueFocus — one of China's largest listed public-relations and advertising groups — co-led a roughly ten-million-yuan seed round in PureblueAI alongside InnoAngel Fund, with Tsinghua's alumni seed fund following 2. BlueFocus called it its first AI strategic investment since committing to an all-in-AI pivot in 2023, but the more candid reading is defensive: BlueFocus sells brand placement and reputation management, the exact function a chatbot threatens to make obsolete when it summarizes an entire product category in one paragraph. Owning a stake in the company writing the new playbook is cheaper than being disrupted by it.

A Venture Firm's Growth Bet

The second check arrived five months later and looked nothing like the first. In March 2026, Vertex China led a tens-of-millions-yuan angel round in PureblueAI, with InnoAngel, Yicun Songling and 36Kr Fund following on 3. This was capital betting on category growth rather than incumbent defense, and PureblueAI used the round to ship its first flagship product: mkter.ai, an AI marketing digital-employee tool called Mark, built on the company's own hybrid model architecture and multi-agent stack 4. Where most GEO vendors sell a single-function dashboard, Mark is pitched as running the whole loop — diagnosing where a brand is missing from AI answers, executing the content and structured-data fixes, and monitoring whether its recommendation rate actually moves. The founding team, led by a former ByteDance Volcano Engine marketing executive who also ran marketing for the Doubao model itself, is stocked with Tsinghua and Chinese Academy of Sciences alumni pulled from ByteDance, Alibaba and Baidu 4 — a roster built specifically to reverse-engineer how those companies' own models rank brands.

From Vendor to Scorekeeper

PureblueAI's more consequential move came in May 2026, when it partnered with 36Kr to publish the second edition of a Consumer Brand AI Recommendation Power Roster at Beijing's AI+ Industry Congress 5. The roster ranks consumer brands by how often and how favorably they surface across major Chinese AI assistants' answers to shopping questions. That is a shift in position: a vendor that also publishes the industry's benchmark scoreboard is no longer just selling optimization services, it is defining what counts as winning. Brands now have a public, repeatable reason to buy PureblueAI's tools — to move up a ranking that PureblueAI itself curates and that 36Kr, one of China's most-read business outlets, distributes to the marketing executives who set optimization budgets.

The Investable Read

For investors, PureblueAI is a useful proxy for a wider bet: that GEO becomes a durable, standalone ad-tech line item in China the way SEO did globally two decades ago, rather than a feature that model owners like ByteDance, Alibaba and DeepSeek absorb into their own advertising products once the category proves its worth. That risk is real and specific. Search engines eventually built their own keyword and ranking tools and squeezed out independent SEO vendors; nothing stops Doubao or Tongyi Qianwen from launching a native brand-visibility dashboard the moment GEO spend gets large enough to notice, especially since the underlying ranking logic sits entirely inside their own models.

PureblueAI's hedge against that outcome is speed and neutrality: two funding rounds and a flagship product inside a year, plus a scoring methodology now attached to a major outlet's brand rather than tied to any single model owner. BlueFocus's participation adds a second layer of protection — it hands PureblueAI distribution into a large existing base of Chinese corporate marketing budgets without building an enterprise sales team from zero. For regional allocators, the signal worth tracking is less any single PureblueAI funding round and more whether other listed marketing and advertising groups across the region follow BlueFocus's lead and take direct stakes in the vendors mediating how AI models talk about their clients. If they do, GEO stops being a startup niche and becomes a line item on ad-tech balance sheets across Asia.