Baidu's ERNIE Consolidation: Too Tidy, Too Late
Baidu has merged its three fragmented ERNIE AI interfaces — ERNIE Bot, Wen Xiaoyan, and Baidu ERNIE Assistant — into a single portal at chat.baidu.com, ending years of confusing brand overlap caused by siloed internal teams. The cleanup is overdue but arrives as Baidu's consumer AI position has already weakened significantly, with ERNIE App at roughly 5 million DAU against ByteDance's Doubao at over 140 million. For PMs, the story is a case study in how organizational structure, not model quality, can determine product fate.
## Why This Matters for PMs The Baidu story forces a specific, uncomfortable question you should be asking about your own product portfolio right now: do your users understand what they're choosing, and if not, is that a marketing problem or a structure problem? Baidu's fragmentation wasn't caused by bad designers or lazy marketers. It was caused by a product organization that allowed multiple teams to ship overlapping products without a forcing function for consolidation. If your company is scaling AI features across multiple teams — and most are, as of mid-2026 — you are at real risk of building your own ERNIE labyrinth. Different teams shipping different AI entry points, each with its own onboarding, its own session state, its own brand language. The concrete action here is a product surface audit. Map every place a user can currently access AI capabilities in your product. Count the login states. Count the chat histories. Count the brand names. If the number is greater than one and the differentiation isn't immediately obvious to a new user, you have a consolidation problem that will only get harder to fix as each team adds more features and more technical debt. The urgency is real: consumer AI habits are forming now. The window to establish a coherent identity in your users' mental model is not permanently open. Baidu had three years to sort this out and didn't. You probably have less time than that.
Baidu merged three ERNIE interfaces into chat.baidu.com, ending years of confusing brand fragmentation caused by siloed business units.
Imagine you're a Chinese internet user in early 2024, curious about this AI assistant everyone keeps talking about. You search for Baidu's offering and find three nearly identical products staring back at you — ERNIE Bot, Wen Xiaoyan, and Baidu ERNIE Assistant — each with its own login, its own chat history, its own subtle branding. You pick one at random, have a decent conversation, then close the tab.
You never come back. ByteDance's Doubao, meanwhile, was right there in the app you already had open.
This week, Baidu finally did what it should have done in 2023: it merged all three of those fragmented web interfaces into a single portal at chat.baidu.com. It's a clean, sensible move. It is also, by most accounts, an admission that the race has already been run.
How Baidu Built a Labyrinth Instead of a Product
The story of ERNIE's fragmentation isn't really a technology story. It's an org chart story. When ChatGPT exploded onto the global scene in late 2022, Baidu moved faster than any other Chinese tech company to respond.
In March 2023, it unveiled ERNIE Bot to enormous fanfare — the first credible Chinese-language AI assistant, from the company that had spent a decade building some of China's most sophisticated natural language research.
But inside Baidu, the excitement triggered a familiar corporate reflex: every major business unit wanted a piece of the action. The search division had its angle. The cloud business had its enterprise ambitions.
The AI research group had its own vision for what the product should be. Rather than forcing alignment, Baidu let each group run with its own version. The result was three products that looked almost identical to users but were organizationally worlds apart — with separate roadmaps, separate teams, and separate brand identities that confused everyone outside the building.
Think of it like a restaurant group that opens three locations on the same block, each with a slightly different name and the same menu, and then wonders why foot traffic is thin. The problem was never the food.
The Numbers Tell a Brutal Story
By mid-2026, ERNIE App — the consumer-facing flagship — sits at roughly 5 million daily active users. That sounds respectable until you hold it next to Doubao, ByteDance's AI assistant, which has crossed 140 million DAU. That's not a gap; it's a different universe.
Baidu was first. It had the brand recognition, the search traffic, the research pedigree. Being first in AI, it turns out, matters a lot less than being coherent.
ByteDance didn't just build a better model — it built a better product experience and distributed it through an ecosystem its users were already living inside. Alibaba did something similar with Qwen, embedding AI deeply into its cloud and commerce infrastructure. Baidu, by contrast, was fighting a three-front war with itself.
The consolidation at chat.baidu.com ends the confusion, but it doesn't conjure back the users who gave up and moved on. Consumer habits in AI are calcifying fast. The people who chose Doubao in 2024 are not comparison-shopping today.
What Baidu Gets Right — and What It Still Has to Solve
To be fair to Baidu, the consolidation isn't purely cosmetic. A single portal means unified login, unified chat history, and a cleaner surface for future feature development. For enterprise customers and developers who use Baidu's AI through its cloud infrastructure, the cleanup probably matters more than the consumer headlines suggest.
Baidu's real remaining strength is in B2B AI — autonomous driving through Apollo, industrial AI applications, and cloud services where ERNIE's capabilities are embedded into workflows rather than competing for casual attention.
The consumer AI market, though, may have already sorted itself out. In China's AI landscape, the pattern looks increasingly like a winner-takes-most dynamic, with Doubao dominant, DeepSeek's models gaining developer mindshare, and a long tail of challengers fighting over the remainder.
Baidu's consolidation is the kind of decision that looks obvious in retrospect and necessary in the present. The harder question — the one Baidu's product leadership is certainly wrestling with — is whether the company's future in AI is as a consumer product at all, or whether it belongs somewhere quieter and more structural, powering the infrastructure that other products run on. That pivot, if it's coming, will be a much bigger story than a URL change.
Frequently Asked Questions
Mostly internal hygiene at this stage. A unified portal reduces friction for new users and simplifies future development, but it does nothing to address the distribution advantages ByteDance has through its existing apps. Users who already have a preferred AI assistant are unlikely to switch because Baidu cleaned up its URL structure.
ByteDance embedded Doubao into its existing content and social ecosystem rather than launching it as a standalone product competing for new attention. It also maintained a single, coherent product identity from launch. Baidu's fragmented approach meant its head start was effectively split three ways, diluting both user acquisition and brand recognition.
It's a broadly common risk. Any large company where multiple divisions are simultaneously building AI features — without a centralized product strategy — can end up with overlapping, confusing offerings. Microsoft has navigated this reasonably well by consolidating under the Copilot brand, though even that has had inconsistencies. The organizational discipline required to prevent fragmentation is harder than the technical work of building the products.
