AutoNavi Pushes Out “DaiMa”: Even a Map Company Is Now Jumping Into Vibe Coding

On June 29, DuJia broke a piece of news that’s not huge but quite interesting: AutoNavi is internally testing a Vibe Coding product called “DaiMa” — a pun on the Chinese word for “code” — positioned as an AI tool that uses natural language to generate WeChat mini programs and iOS native apps with one click.

A company that makes maps and navigation suddenly comes out with an AI programming tool — it feels like a big leap. But if you’ve been following Alibaba’s moves lately, you’ll see it actually makes perfect sense.

What Exactly Is DaiMa?

Let’s clarify the product first. DaiMa’s core selling point is similar to many existing Vibe Coding tools: you don’t need to know code, set up a development environment, or be technical at all. You just type your idea in plain language, and AI spits out a runnable application.

But there are a few details worth highlighting:

First, the output is a “launchable” product, not just a demo. This is where DaiMa sets itself apart from many toy-like tools. Overseas products like Bolt.new or v0 mostly stay at generating Web pages or React components, and if you want to get something into the App Store or the WeChat Mini Program backend, there’s still a lot of tedious engineering work involved. DaiMa claims that once generation is done, it gives you a QR code for a real-device experience — scan it with WeChat, and it runs on your phone. If this closed loop actually works, it could save independent developers a ton of time.

Second, the platform choice is pragmatic: WeChat Mini Programs + iOS native. No Android, no Web, no chasing cross-platform hype. The common feature of these two platforms is that they are the largest domestic traffic pools for truly “lightweight commercial apps.” If a business owner wants to make a membership booking tool, product showcase, or internal tool, they either go with a WeChat Mini Program or build an iOS app. DaiMa’s scenario targeting is laser-focused — it’s aimed right at small merchants and independent devs.

Third, iteration is done via “conversational refactoring.” If you’re not satisfied with the initial output, you keep describing changes in natural language, and AI updates it automatically. Unlike Cursor’s “AI writes code, you review” model, DaiMa goes for a truly No-Code approach — users never see the code and don’t need to.

This product decision is actually quite bold. Cursor, Windsurf, Claude Code — these all treat the user as “someone who knows code” and position AI as a copilot; DaiMa takes the wheel entirely, with users just deciding the destination. The tradeoff? Low ceiling for customization — complex logic is unlikely to be supported. But for its target audience — small business owners, product managers, freelancers who don’t code — that’s exactly what they want.

Why AutoNavi?

This is an even more interesting question.

AutoNavi’s role in the Alibaba ecosystem has been changing over the past two years. From pure map navigation, to a local life entrance, to now trying to make developer tools. Path-wise, it looks like a leap; logically, it’s actually a straight line: AutoNavi has the most complete LBS data and local merchant relationships, but lacks a developer ecosystem to “put these resources to use.”

For example: a restaurant owner wants a membership mini program that can locate nearby users, push coupons, and link to their AutoNavi store info — traditionally they’d hire an outsourcing company, spend tens of thousands of RMB, and wait two months, only to end up with something that might not work well. If DaiMa allows them to describe their needs, generate in half an hour, and launch the same day, AutoNavi effectively upgrades local merchants from “information display clients” to “app producers.”

This is a microcosm of Alibaba’s AI strategy: the Tongyi large model as the base, each business unit finding its own way to apply it. Taobao does AI shopping guidance, DingTalk does AI office, AutoNavi does AI programming — everyone wraps model capabilities into products for their own familiar scenarios.

Timing-wise, AutoNavi’s entry isn’t too late. Vibe Coding was coined by Andrej Karpathy in early 2025, and in the year since, ByteDance’s Doubao MarsCode, Tencent’s CodeBuddy, and Alibaba’s Tongyi Lingma have all entered — but these target programmers. For a pure beginner audience with complete app generation in a No-Code approach, there aren’t many solid players in China. Mendix and OutSystems’ enterprise-level low-code is too heavy for the Chinese small business market; domestic low-code platforms mostly still rely on drag-and-drop, with insufficient AI integration.

If DaiMa succeeds, it could fill a pretty empty niche.

A Few Unavoidable Questions

Having described the pros, now let’s nitpick.

Where’s the ceiling for output quality? This is the Achilles’ heel for all Vibe Coding products. WeChat Mini Programs are manageable due to their restricted environment and standardized components; iOS native apps are much harder — AI has to handle Swift, UIKit/SwiftUI, various API permissions, code signing and packaging, and review processes, each with its own pitfalls. There’s a Mariana Trench between “usable on a real device” and “approved for App Store.” DaiMa hasn’t disclosed how it solves review issues — likely something they’ll hit trouble with later.

What about complex requirements? Natural language descriptions are great for simple scenarios — “make a bookkeeping mini program that can take a photo of receipts” is within AI’s reach. But if a user says, “I want a membership system with three-tier distribution rebates, integrated with enterprise WeChat SCRM, tied to ERP inventory,” AI will likely get confused. If DaiMa truly focuses on lightweight scenarios, its growth ceiling may be lower than expected; if it aims for complex scenarios, it will grapple with the hard problems Cursor and others are already tackling.

No mention of business model. This is intentionally omitted in the official announcement — launch time, industry partnerships, pricing — none disclosed. Possible models include: pay-per-generation, tiered pricing based on app complexity, bundling with Alibaba Cloud compute resources, or free generation with LBS API fees for merchants. Which works will depend on beta feedback.

Trade-offs in ecosystem binding. Apps generated by DaiMa will almost certainly be tightly integrated with Alibaba Cloud, AutoNavi SDK, and Tongyi large model services. For small merchants, that’s fine — they don’t care about tech stack purity; but independent developers with ambitions may hesitate over being locked into a closed ecosystem.

The Vibe Coding Track Will Be Lively This Year

Looking from afar, the Vibe Coding track in the first half of 2026 is evolving faster than expected.

• Overseas, Anthropic has made Claude Code a first-class citizen, and Cursor’s valuation reportedly nears $10 billion USD
• In China, ByteDance, Tencent, and Alibaba are already in, now joined by AutoNavi
• Model capabilities-wise, Claude 4, GPT, and Gemini series have closed the gap in code generation — the remaining competition is in engineering and product form

In short, the model-layer dividend is mostly eaten — the next wave is about “who can turn AI coding into a product that ordinary people can use.” AutoNavi’s choice to start with small merchants creates a differentiated competition against ByteDance/Tencent’s developer-focused approach — a well-placed first move.

As a side note, developers can now use aggregation API platforms like OpenAI Hub to call Claude, GPT, Gemini, DeepSeek, etc., with a single API key, comparing code generation results without registering and recharging each separately — a real time-saver for teams building Vibe Coding tools or their own AI coding assistants.

As for how DaiMa will ultimately perform — we’ll have to wait for public access to test real output quality. Based on official descriptions, the specs look good on paper, but Vibe Coding products often wow in demos and disappoint in real use. 90% of the user experience comes from the last 10% of detail polish — error handling, debug feedback, breaking down complex requirements — each is ten times harder than generating a “Hello World” app.

Once AutoNavi announces launch dates and first beta channels, we’ll do a hands-on evaluation.

References

• ITHome: AutoNavi Reportedly Testing “DaiMa” Vibe Coding Product — First report on DaiMa’s positioning, use cases, and target users