Reporter |
Editor | Wen Shuqi
On June 10th, we got the exclusive scoop that Dreame’s ecosystem brand, Eclix, has teamed up with a leading Chinese big model company. They’re planning to drop their first-gen AI phone by September 2026, just in time for the Double Eleven shopping spree. This ain’t your average smartphone – it’s a premium beast, priced over 5,000 yuan (about $700). The core team is basically done with R&D and is gearing up for full device testing.
The Eclix phone squad is put together by Yu Lei, a former VP at Gionee. He officially joined MOVA back in March 2026 to head up their phone and AI hardware gigs.
Yu Lei is a total veteran in the phone game. Back in 2015, he hopped on board with Gionee as their Global VP, running marketing and channel sales, and even led a major brand makeover. Fun fact: He’s also launched a smartphone brand before, focusing on overseas markets.
We heard straight from Yu Lei that Eclix will live on as a Dreame ecosystem brand, separate from MOVA, with a crew of about 60. Most of the core members come from AI or OS backgrounds, and they’re looking to scale up to around 300 folks by the end of 2026. Dreame is incubating the Eclix team, but they’re also open to outside funding. The AI phone project is already chatting with several investors.
Yu Lei told us they’re keeping the first-gen production tight – just 50,000 to 100,000 units – launching exclusively on JD.com. The target audience? Hardcore tech geeks and digital enthusiasts. “We’re not trying to sell a million units out of the gate,” he says. “We want to use small batch deliveries to polish the product and lay a solid foundation for future versions.” He also mentioned that thanks to Dreame’s existing distribution channels, they’ve already racked up orders for over 100,000 units.
The Revolution: No More Apps
“Our hiring vibe can be summed up as ‘rebels’ – people who come from big companies, have solid skills, but are bored of just doing one thing and want to dive headfirst into a new product vision. This kind of talent keeps the team super agile and innovative. Honestly, when I was younger, working at a foreign company, I was one of those typical big company rebels,” Yu Lei shared.
From day one, Yu Lei has been pouring a ton of energy into recruiting. He admits it’s tough finding the right AI phone talent these days. Before Toutiao’s ‘Doubao’ phone project, similar roles paid a few hundred thousand yuan a year. Now, the market rate can hit 5 to 8 million yuan, but the Eclix team is making sure they’re paying at least as well as their last gig.
“When I told the model companies and my team I wanted to build a phone with zero apps, everyone got super excited. That’s how we attracted so much top talent,” Yu Lei told us.
We’ve learned that the Eclix phone project is called Project Quasar internally. It’s built on an AI Agent core OS, using a full-stack, self-developed approach that blends hardware, software, and OS. This enables a whole new way of interacting – driven by natural language, multi-modal input, and intent prediction. It’s a total departure from today’s AI phones that just simulate clicks on existing apps or act as a glorified voice assistant. This thing actually gets tasks done without any simulation.
The secret sauce of the Eclix AI phone is replacing apps with “atomized services.” For example, just say, “Order me a 25-yuan grilled chicken burger,” and the system will directly tap into integrated life service providers to place the order, auto-filling your preferred address and taste. All you have to do is confirm. No app to open, no menus to navigate.
Yu Lei revealed that Eclix is already working with a top big model company to develop and test the whole pipeline – from voice commands to executing atomized services. They’re focusing on high-frequency scenarios like food and entertainment first. He believes that to make a great AI phone right now, you absolutely need the support of a big model company. Partnering with one is the best way to get the ball rolling.
Gu Wei, Eclix AI’s Chief Technology Expert, thinks most “AI phones” on the market today are just regular phones with a big model slapped on top. Open them up, and you still see the same old grid of app icons.
The core change with the Eclix AI phone is the interface: users won’t see any traditional app entrances. The AI is the main gateway, and most actions are done through voice commands. The system directly calls atomized services based on your instructions, no hopping between apps required.
Unlike “screen-reading” control modes, Eclix uses a direct cloud-service connection. It doesn’t need to read your screen or manipulate your apps, which sidesteps a lot of potential compliance headaches.
“We classify risk levels as low, medium, and high, following current MIIT regulations. Medium and high-risk actions need explicit user approval. All operations are logged and traceable, and you can undo or revert any action. Compared to ‘phone + AI assistant’ type AI phones, our next-gen AI phone is more like the jump from feature phones to smartphones. It’s not a gradual upgrade; it’s a total ground-up rebuild,” Gu Wei explained.
According to Gu Wei, the Eclix AI phone uses a device-cloud collaborative architecture they call the “Dual-Brain Architecture.” Simply put, it splits user needs into two categories: one for tasks that need to be fast, accurate, and controllable (execution tasks), and another for conversations that require understanding, companionship, and long-term memory (dialogue tasks).
Breaking it down, the “Left Brain” handles execution tasks like ordering, searching, settings, and calling app functions. It’s all about accuracy, stability, and speed – getting the deterministic stuff right. The “Right Brain” manages long-term dialogue, emotional companionship, and multi-modal interaction. It focuses on your state, preferences, and memory, making the AI feel like a true personal assistant who’s always there for you.
Gu Wei further explained that the reason for the “Dual-Brain” is that a single model struggles to be “fast, accurate, and deep” all at once. When you push for speed, complex understanding suffers. When you aim for deep experience, response time and certainty take a hit.
The beauty of this approach is separating rational execution from emotional companionship. Then, through engineering-level scheduling, verification, and permission controls, they can reduce the impact of big model hallucinations on critical tasks. This lets the AIOS Agent both chat and actually get things done for you.
Plus, with your permission, long-term memories are stored in your own account on the project’s private cloud. So, no matter which underlying big model you switch to, your memories – your preferences, habits, and all – are safe. The AI will get smarter about you over time. It might be a bit clumsy at first, but as data accumulates and services connect, the smarts will keep getting better.
We’ve learned that the “Dual-Brain Architecture” is a key part of Eclix AI’s AIOS Agent design, and they’ve already filed for patents on it.
Open Ecosystem, No Hardware Profits
Yu Lei told us that the big model company Eclix is partnered with is bringing its own ecosystem resources to the table, and they’ve already integrated multiple applications. The core revenue model? Subscriptions for AI computing power services. They’re borrowing from the existing playbook of monthly AI model subscriptions and token-based billing. Users will be charged in real-time based on the tokens consumed by different models, similar to how mobile data is billed. You just pre-pay for credits.
In his view, this model lets them keep hardware margins razor-thin to compete with traditional phone makers. Big traditional phone companies make over 8 billion yuan annually just from app distribution, with crazy-high margins. Switching to a new model means admitting their old success path is dead, and it’s not a road they’re likely to take willingly.
He compares the current industry stage to the eve of the smartphone explosion in 2006-2007. Back then, Android had just 6% market share; four years later, it owned 80% of the non-iOS market. The parallel is strong: the market is far from settled, AI phones are at a tipping point, and there’s still room for a new giant to emerge.
“Big companies aren’t incapable of doing this; they’re just afraid to. They make a killing on app fees. We’re going to kill the app, starting by atomizing its capabilities. In the future, apps might not even exist,” Yu Lei stated firmly.
The AI phone buzz is getting real. Besides Apple, Samsung, Xiaomi, and Honor ramping up their AI efforts, ByteDance is expected to release its second-gen ‘Doubao’ phone this year. We’ve heard from supply chain sources that the new Doubao phone is prepping over ten times the volume of the first gen – hundreds of thousands of units. OpenAI and Jieyue Xingchen have also announced plans to launch AI phones this year.
Rumors are flying that OpenAI’s plan is to build a screen-less phone. Yu Lei told us both brands are disruptors in the phone game, and the core advantage lies in their team’s depth of thinking and execution. Eclix’s approach is similar to OpenAI’s, but they’ll keep the screen for now – though they’re not ruling out other form factors later. As for familiarity with the phone supply chain, Yu Lei says Eclix is way more confident.
On the ecosystem front, Eclix is staying open on both hardware and AI-native software. On the hardware side, they’ve already connected with multiple BU’s within the Dreame ecosystem, including cars. On the software side, they’re pushing “No App” by using open tools and data access via MCP, letting third-party apps, smart devices, and enterprise systems plug in. They’re also supporting ACP for external AI agents, hoping to co-build a new AI-native software ecosystem with app developers.
Yu Lei shared that the big model companies in China are generally on board with this direction and are keen to collaborate deeply. The stars are aligning, and the window of opportunity is open. But he admits they’ll need to make some compromises for now. The biggest one is what they call the “escape window.” They can’t integrate every app yet, so they’ll still leave a “legacy escape window” for super apps like WeChat and banking apps.
“Rome wasn’t built in a day. We can’t convince WeChat and other super apps to natively integrate right now, and we’re not going to force it. But we’ve kept a compatible plan. The first-gen product will have traditional apps as a safety net, but that’s a leftover from the old world,” Yu Lei explained.
We’ve learned that sales channels for the first Eclix AI phone are already set up. Plus, their other project, the ACKA outdoor sports phone, is about to launch in several global markets. They’ve already received orders and deposits for millions of units, which will provide solid operating cash flow. Eclix estimates the team will turn profitable by Q4 2026.
