Ant Group should acquire Macaron.

"The most founder-respecting episode ever 😭"

After trying Lingguang, I've made peace with Macaron.

I still think Macaron's thesis is flawed, but that doesn't matter. The art of storytelling is unfalsifiability — as long as Macaron can keep its narrative coherent and keep extracting gold coins from investors, that's enough.

Spending VC money to bring joy to the AI industry — that's Macaron's romanticism 😭

Unfortunately, during this period I came across Lingguang, an app from Ant Group that bears a striking resemblance to Macaron.

Lingguang's biggest problem is that it has no thesis. A major internet giant like Ant Group built a budget Macaron 😭

It's been over ten days since I downloaded Lingguang. I wanted to roast it from day one, but I've had nothing to say.

Speechless, basically. This thing is too boring.

Lingguang is essentially the boomer version of Macaron — slapping some frontend-generation gimmicks onto a chatbot.

Lingguang has two main features. First, it has decent layout capabilities: when answering questions, it generates images and inspirational quotes.

I asked it, "How did the Catholic Church respond after the fall of the Western Roman Empire?" Lingguang's response wasn't just text — it generated two images, two tables, and capped it off with an inspirational quote:

"When the empire lies in ruins, the lighthouse of faith still stands, illuminating Europe's Middle Ages. — Historical observation"

I laughed when I saw this. Can anyone explain what user need this solves? What does adding two images and an inspirational quote actually accomplish?

I'm deceased — Ant doesn't actually think these gimmicks can compete with Doubao, does it 🥹

Lingguang's other flagship feature is "Flash Apps" — generating mini-programs with only frontend interfaces.

I said, "Create a schedule assistant to remind me to wake up at 8 AM tomorrow," and it generated a frontend mini-program.

But here's the funny part: Lingguang doesn't have notification permissions.

I asked Lingguang to remind me to wake up at 8 AM tomorrow. It first generated a futuristic alarm clock image. Then told me "Wake-up reminder set," and confirmed the reminder time one more time. Finally, it quoted the proverb, "The early bird catches the worm; the early riser catches good health."

Pretty funny.

Very structured response. Very correct proverb.

But Lingguang doesn't have system notification permissions. It didn't even notify me within the Lingguang app itself. Very good Lingguang, makes my brain spin 🧠

My verdict: Lingguang is genuinely worse than Macaron.

Because I tested the same case on Macaron. Macaron can already access system notification permissions. When I created a schedule reminder assistant on Macaron, it actually popped up a notification to remind me.

The only issue was that the displayed time was wrong — I set an afternoon alarm, but it still showed "It's 8 AM now, time to wake up." 🤣

I even reported this bug to Kaijie, and he said he was going to beat up Macaron. I don't know if he meant beating up the programmer or the server — I'm deceased, he really 😭

Setting aside this minor detail, Macaron is at least usable now.

But Lingguang? My friend, released four months after Macaron, a product from a major internet company, with executives saying they mobilized a 200-person "AGI Exploration Group" to develop it.

I don't understand (in Zhejiang Fenghua accent).

I'm not even talking about thesis — just product completeness. Lingguang, with several times the resources, is less interesting than the hype product Macaron 😭

Sorry Kaijie, I was wrong. I used to think you were all hype, but now I see you actually have innovation, resilience, and execution speed that leaves big companies in the dust.

Of course, Lingguang isn't without merits.

Lingguang's advantage is speed — generating a mini-program takes just dozens of seconds, and opening already-generated apps has virtually no latency. Macaron has never optimized this issue; maybe Kaijie spends all day beating people up and the devs are too traumatized to function.

But what can I say?

This engineering advantage is Lingguang's birthright. An app made by Ant Group — if opening a built-in mini-program took a minute, that would genuinely be disrespecting the payroll.

Back to the core thesis issue.

I still believe that Macaron's story of "ordinary users hand-crafting AI apps, democratizing technology to serve long-tail needs" has a fundamental problem: ordinary users lack product-definition capability.

This story is too aimed at product managers, too aimed at investors.

My first reaction to these Vibe Coding communities, these Personal Agent apps, is: I don't know what to do with them.

All my daily needs are already solved by mature apps — ride-hailing with DiDi, maps with Amap, reviews on Xiaohongshu and Dianping, schedule assistant with Toki — which lets you set a shortcut command, tap the back of your iPhone twice, screenshot and send to Toki, then it recognizes the screen and sets the corresponding schedule reminder.

This kind of useful small feature definitely has no technical barrier; what it requires is developers who can identify and define user needs. Ordinary users lack this ability.

Four months after Macaron's launch, its inspiration library aka Agent store still has the same two pinned mini-programs: photo-to-3D and photo background replacement.

This is entirely because the Nano Banana model is impressive — it has nothing to do with this story about ordinary users hand-crafting apps.

I still think Vibe Coding is a developer-facing thing. Democratizing technology means lowering the barrier to building apps from programmers to professionals — doctors, lawyers, directors, writers, and pig farmers — who can develop their own workflows.

Building an Agent store, a Vibe Coding community, first requires finding this cohort of technically-inclined professionals and figuring out how to incentivize them to build small products.

Rather than directly telling a personal assistant story aimed at ordinary users.

Of course, as I said at the beginning, the art of storytelling is unfalsifiability. Macaron's narrative is indeed coherent, Kaijie does have innovative ability, and Macaron's design aesthetics are cliff-leading ahead of the AI product pack.

So here's the question: why doesn't Ant Group just acquire Macaron?

Ant Group has money, engineering capability, user base, and can pour unlimited ad spend. These are things Macaron doesn't have — maybe only in Kaijie's dreams.

Macaron has aesthetics, innovation, narrative ability. In short, it has some life. And that's exactly the vitality that big companies lack.

So, I hereby sincerely recommend that Ant Group acquire Macaron.

Macaron is valued at under $100 million, and with 20-something people pulled off something more interesting than Ant's 200-plus team.

Acquiring Macaron would definitely be more cost-effective than investing in Lovart at a $500-600 million valuation 😭

(Images in this article generated by Gemini, writing assisted by Claude Code)

Livestream preview for tomorrow night:

I found a friend who's been intensely using the Doubao phone these past two days — Zehao Yang, a blond guy building his own Agent phone. He candidly and clearly told me that his Agent phone got killed 👍

Tomorrow at 8 PM, we're going live to discuss whether the Doubao phone is legit.