We attended this year's MiraclePlus Demo Day, and we think only these seven AI projects are the future.

AI闹·June 11, 2025·0·0

Spring 2025 Demo Day at the MiraclePlus Forum. Fifty-seven startups on stage, and the thousand-seat auditorium at the Zhongguancun International Innovation Forum was packed to capacity.

Produced by AI NOW! 01

Spring 2025. Demo Day at the Qi Ji Forum. Fifty-seven projects. The Zhongguancun International Innovation Forum's thousand-seat auditorium packed to capacity.

Reportedly the most crowded year in Qi Ji history. Investors showed up nearly in full force.

If you weren't there, you might want to ask yourself why.

Either you're high enough on the ladder that you're used to letting your team handle these things for you.

Or you've mentally "checked out" of the scene, your information anxiety no longer acute enough.

Or you're a peripheral player, and your presence or absence doesn't really move the needle.

Whatever the case, this was the event to be at this year. Missing it was the wrong call.

Qi Lu delivers his opening remarks, sharing his latest thinking on AI for the first half of 2025

Overview of Qi Ji's 2025 cohort

Yiming Zhang spotted in the audience

As for the projects themselves, the chatter on the floor was telling.

"Not as good as last year."

"That garbage project we passed on in '22? Somehow it's actually working now."

This is precisely the honest portrait of AI entrepreneurship right now. This isn't a static comparison game. It's a dynamic long-distance run.

What's AI NOW!'s take?

Qi Lu put it well — this is the -1 to 0 stage, not even 0 to 1.

In this interval, projects are naturally chaotic, naturally half-baked.

Anyone insisting on PMF or ARR in this kind of environment is more likely to misread the field.

So what should you look at?

It's not simply "bet on the person." Founder charisma matters, but it's far from sufficient.

We suggest a different framework.

Among these 57 projects, which ones are worth betting on, worth running alongside, worth at least knowing the founder?

Breaking them down across these six dimensions might prove more useful.

1. Build Power

Shipping beats talking.

The ability to build products and engineer solutions is the hardest moat in AI entrepreneurship.

Whether it's a billion-agent network or a multi-peptide drug discovery platform, build or die.

2. System Sense

Does the team think in systems? Do they grasp industrial chains, ecosystems, future interaction paradigms?

It's not about single-point tech or features. You need to understand industrial transformation, human nature, and network effects.

3. Temporal Sweet Spot

Are the founders at the golden intersection of personal trajectory, technological moment, and industry inflection?

Age isn't the point. Timing perception and execution are.

Dropout or PhD, Big Tech veteran or never worked a day — what matters is making the right move at the right moment.

4. Tech Shift Potential

Is there a chance to drive industry-level technological transformation, or introduce new interaction paradigms and behavioral shifts?

Not "a better tool," but "rewriting the rules of the game."

5. Early GTM Clarity

Is there a clear and plausible early market entry path? Who's the customer, why will they buy, how will they use it?

Not "we believe there's a market," but "we know how to open the board."

6. Founder Drive & Gravity

Does the founder possess extraordinary drive, the kind that attracts resources, talent, and capital to orbit around them?

Especially critical in AI Native tracks — can the founder craft narrative plus community, generate momentum before product-market fit?

Using these dimensions, AI NOW! picked its 7 favorite projects. In our view, only these —

are the future.

Listed in no particular order

Magic Gourd

Founder: Liwei Wu. A "second brain" tool for intent prediction, moving beyond prompt-driven interfaces.

Already launched, cross-platform capabilities preliminarily validated, strong build power. System design leapfrogs from "prompt-driven" directly to perception-based interaction via photo and voice recording — a clear paradigm shift. Timing is precise, catching the interface moment between personal AI and sensing hardware. Tech shift direction is clear, though the path carries some dependency on hardware ecosystem. Early community users exist, with use cases still expanding. Founder comes from SenseTime's IPO core team, execution is steady, team chemistry is high.

Build Power: 5 | System Sense: 5 | Temporal Sweet Spot: 5

Tech Shift Potential: 4 | Early GTM Clarity: 4 | Founder Gravity: 4

Mirror Bloom

Founder: Zhouyingcheng Liao. Intelligent 3D character animation engine — create your own Toy Story.

Core technology refined, demo clear, product at critical juncture for commercialization. Team demonstrates systemic understanding of virtual being ecosystems and 3D content production pipelines, entry timing aligned with AIGC's evolution toward 3D. High tech shift potential, with prospect of triggering content interaction paradigm change. Current product scenarios are clear, market thesis still being continuously validated. Founder is a graphics PhD who left school to start up, strong research background (multiple highly-cited top conference papers), with artistic temperament — former keyboardist in a metal band who went on tour. Complementary team configuration, strong execution.

AILSI

Founder: Serena Yang. End-to-end AI multi-peptide drug discovery platform.

Young female founder with dual academic and industry background, compelling presence on stage. Platform loop already built, actual orders landed, demonstrating excellent build and execution capability. Deep understanding of biopharma industrial chain, combining technical landing with industrial integration system sense. AI+biopharma window remains open, though industry development is rapid and speed of positioning is critical. Technology has potential to transform drug development workflow, already secured orders from top-tier pharma companies, commercialization path is clear.

Bonjour

Founder: Vincent (Zheng Xie). The talent marketplace for the Agent era.

Charismatic Gen Z founder born in 2002, with extraordinary personal drive and reality distortion field, known in his community as the "Liangzhu Steve Jobs." First-generation product already launched, fast validation, substantial user numbers, strong community cultural stickiness. Deep understanding of Agent-driven interaction logic, clear boundaries. Agents are at early stage of cognitive diffusion — timing is appropriate but carries risk of being too early. Overall project demonstrates clear tech shift trend and early market response.

Infotopia / Nanqiang Beidiao

Founder: Liwen Zhang. Gen Z Agent voice community.

Product not yet officially launched, Xiaohongshu side has preliminarily completed demand validation. Deep insight into young people's desire to "be understood," clear product loop conception, novel entry point combining Agent with audio medium. Current window is suitable for experimentation with audio expression Agents, but community cold start and retention remain challenges. Founding team has distinctive temperament, skilled at building emotional resonance with Gen Z.

According.Work

Founder: Qinghao Huang. A "company-of-one" collaboration network — so no one has to be a cog anymore.

Strong team execution, delivered 500+ open-source issues in six months, nearly a thousand developers involved. High system design level, structural thinking evident from organizational form to technical mechanism. Cuts into the structural crack between "individual vs. organization," timing is sound. Future potential even greater if collaboration mechanisms can expand beyond developer communities. Founder has rich background, combining engineering power with narrative power, former murder mystery game entrepreneur, carries clear "atypical builder" temperament.

Vector Story / Loud Chapters

Founder: Tianquan Zhao. Building The Walt Disney Company of AI-era content production.

Founder Tianquan Zhao combines platform operator and content creator experience, clear communicator, well-matched team composition. Product system covers three ends: courses, writing tools, and works, fast landing rhythm. Timing precisely chosen, AI writing boom叠加content platform gap creating opportunity. Platform attributes still in early construction phase, tech shift currently driven by workflow restructuring, not yet broken through圈层. Already has paying student monetization and chart-topping work cases, initial validation is persuasive.

Finally, we think the scores don't matter. Who's judging you doesn't matter.

If you're certain that what you're building isn't a safe project but a disruptive one;

If your goal isn't "filling a demand gap" but shaping an entirely new context;

If your product isn't "a better tool" but something birthing new behaviors and organizational forms;

If you aren't "an excellent operator" but a worldview-carrying detonator.

Then traditional VC frameworks may not be able to measure you.

But we can feel it — they're definitely the sharp edges emerging from this new generation of founders.

That's enough.

Images | Shot on site by AI NOW!