智东西（公众号：zhidxcom）
作者 | 云鹏
编辑｜漠影
现在，试想这样一些场景：
当你上课时遇到没听懂的地方，轻点两下夹在领口上的录音豆，课后AI就会从你点击的那一刻讲起，为你细心答疑，不需要举手、不需要打断任何人；
当快递件丢失时，AI系统直接通过摄像头逐帧分析，智能锁定可疑人员，快速定位，输出一条拼接完整的视频，省去大量人工复查时间、大幅降低沟通成本；
在大洋彼岸的欧洲用户家中，家庭储能智能调度系统通过AI优化光储经济，减少太阳能浪费，一年为用户省下数百欧元的电费；
今天，这些场景绝非空谈，而是选手们在为期24小时的安克AI黑客松大赛中做出的优秀作品。这些作品有着成熟的方案构想，甚至基于真实的安克智能硬件产品，在大赛现场真正“跑起来”了。
场景与作品的背后，是一场大比拼：来自180多所高校和开发者社区的1800多份报名历经层层选拔，最终33支队伍杀入线下决赛，在24小时内完成了从想法到可演示、可验证成果的转换，4支冠军队伍脱颖而出。
放在过去，24小时做出这样一个能跑起来的版本，本身就是黑客松最大的看点。但今天，在AI的辅助加持下，“快速写出一个Demo”的成本趋近于零。Stack Overflow 2025年开发者调查显示，84%的开发者正在使用或计划使用AI工具。黑客松真正稀缺的已经不再是代码能力，而是值得解的真问题、能接入真实世界的接口，以及作品做完之后的去处。
安克的首次黑客松大赛，提供了一个可观察的新样本。
一、 24小时的价值正在被重新定义
“互联网改变的是距离，AI时代改变的是时间。”这是安克创新CIO Neil对AI的判断。
这句话，安克自己先验证了一遍。安克创新2026年半年度报告显示，过去一年公司总Token消耗达30万亿，员工日均Token消耗4500亿，同比增长30倍；AI承担了超过95%的基础代码生成，节省约80%的开发工时。
在黑客松大赛的Workshop分享中，我们看到创客们可以用Launch平台快速上线路演作品，作品可以直接转化为可访问链接，从硬件360度拆解展示、产品使用模拟动画、像素风交互演示到对接动态数据源生成实时更新的BI看板，平台功能应有尽有。
这样优秀的产品源于安克内部的“AI基础设施”高速迭代：基于AI的辅助开发，25年年中，团队仅用数天就完成了第一版原型；今年清明节假期期间，团队仅用3天完成全功能上线，在全公司正式推广。从数天到3天，被压缩的正是Neil说的“时间”。
放在24小时的黑客松赛制里，这意味着：写代码的时间被AI压缩之后，选手之间拉开差距的，不再是谁敲得更快，而是代码之外的东西——创意可不可行、使用场景选得对不对、用户痛点和问题抓得准不准。
正如大模型狂飙突进之下，行业核心关注焦点已经转向AI能否在真实世界中落地应用、实现商业价值。
安克黑客松向选手们开放了真实场景、真实问题、真实产品SDK，让诸多优秀创意和作品获得真实生长土壤，这背后，是安克对AI的长期坚定投入，以及对于极致创新的积极探求。
那么，代码之外真正拉开差距的“真问题”和“真实接口”到底长什么样？我们不妨来关注这场黑客松的冠军作品。
二、“拆解”决赛落地作品，理想如何照进现实？
要看懂这些冠军作品为什么能赢，先要换一把尺子。在WaytoAGI创始人AJ看来，模型决定AI的上限，而场景决定AI的价值。换句话说，代码能力如今人人都有，真正拉开差距的，是选手有没有找准“值得解的真问题”，有没有把它放进“真实世界里能跑通的方案”。
▲WaytoAGI创始人AJ
这场比赛的打分标准也印证了这点。产品创新与AI原生性占比达到35%，用户价值与场景落地同样占比35%，两者相加近七成。最终脱颖而出、能让人记住的，恰恰是把“真实”两个字做实的作品。它能创新性地解决真实需求痛点，拥有落地价值。
为期2天的活动中，我们身在现场，看到创客们激烈探讨、灵感碰撞、熬夜奋战开发。最终的路演环节，智能录音、智能安防、充电储能、智能服务四大赛道上都涌现出诸多令人印象深刻的作品。
在智能录音赛道，“举手豆”团队利用安克Soundcore录音豆的功能以及背后AI大模型的上下文理解、记忆、个性化服务能力，打造了一套独特的AI辅助课堂学习方案。
团队成员在实际的生活学习中发现，很多学生因为担心打扰他人，羞于主动举手提问，这影响了学生对课堂内容的消化吸收。
为此，他们利用AI技术和智能录音设备，实现针对性的AI讲解答疑，提升学习效率和教学质量。同时，这款产品还能让老师更好地了解学生学习状态，得到教学效果的及时反馈与分析，进一步辅助优化教学策略。
从真实的教学痛点、真实的录音豆产品到“举手豆”的诞生，在AI智能硬件和AI技术加持下，团队将想法变为现实。
评委中的安克录音豆团队相关负责人在现场点评环节特别提到，这一作品所解决的痛点与他们实际研发中的功能高度一致。这意味着作品方向确实可行，未来并有望落地验证。
在安克海外市场长期深耕的智能安防领域，“驿站追迹”团队将目光放到了“快递丢失”这一棘手物流问题上。
团队利用Anker eufy智能摄像头、开放的产品SDK以及AI技术，通过对多路视频流进行智能分析，实现了对丢失快递的高效识别和追踪，解决了传统物流监控中人工审核低效且不准确的痛点。
团队成员特别提到，经过测试，他们发现安克的摄像头有着安装简易、功耗低的特点。在实际应用中，安克设备对人物的动态追踪和时间关键词提取功能效果最符合需求。团队利用安克产品已有的功能，简化了整个产品的落地流程，并解决了低功耗和多路并发的问题。
正如前文所说，安克海量产品在实际消费者使用中，能够收获大量真实反馈。这些反馈可以给智能服务类产品的打造提供真实的数据土壤。
智能服务赛道的冠军OntoCare本体队，通过构建服务本体和决策本体，以图谱节点方式增加业务规则和SKU，实现SOP快速更新迭代，相比传统Workflow和Prompt方案，面对多国政策法规变化时，可以更好规避底层代码重构问题。
此外，团队还通过红蓝对抗和案例驱动，开发了一套自进化系统，将客服问题解决率从83%提升至98%。
在赛后采访中我们了解到，这一项目的灵感来源于安克实际客服场景面临的挑战，团队从真实客服系统中找到痛点，在真实售后数据中得以检验。
在一个个优秀团队案例中我们可以看到，基于AI技术以及安克的智能硬件、SDK等底层支撑，Builder能够快速将自己的想法变为现实。更重要的是，这些想法源于真实场景中的真实问题，有机会演变为千万用户用得上的技术。
在今天的AI创业浪潮中，最大的陷阱往往是“拿着锤子找钉子”：先有AI技术，再去生搬硬套应用场景。而安克黑客松上的优秀作品，则是基于安克亿级的设备数，这些应用一旦落地，就可以在真实场景中形成数据飞轮，进一步加速AI技术的迭代，这也是安克在AI时代展现出的巨大潜力所在。
三、 让想象落地，给灵感插上“真实”的翅膀，安克AI黑客松为何如此难得？
为什么上述这些优秀的创客作品我们很难在其他黑客松赛事中看到？这就不得不提到安克AI黑客松的“稀缺”之处，源于四个关键方面：
第一，是从赛题、场景、产品、技术到最终评价标准都紧密围绕“真实”二字。
此次四条赛道的赛题都源于安克真实产品线与真实业务需求。而智能录音、智能安防、充电储能、智能客服这些领域几乎都有一个共同特点：绝非短时间内可以做好的赛道，都需要长期大量投入、扎实的产品布局、实打实的市场份额和用户。
这些不是纸面上的赛题，而是安克创新做了十几年、全球几千万家庭正在使用、已经进入市场的真实产品，是有源之水、有本之木。
在AI已经足够强大、足够普及、足够能干活的今天，竞争焦点已经从技术转向了场景。
第二个稀缺之处，是敢把“家底”开放给参赛者的底气。
安克敢于将正在服务全球上亿用户的接口向本次黑客松的参赛者开放，展现了公司强大的技术自信。
这些SDK对Builder们来说是真正好用、易用、实用的SDK，这些关键能力通过SDK被快速调用，让很多想法在真实产品上得以实现。这离不开安克底层技术接口极高的完备程度和安全性。
参赛者们并非从零开始，而是站在成熟的技术和产品底座上创新。纵观行业，安克敢于在赛事中开放底座的这份勇气，无疑能让这家公司更好地吸引全球创新人才。
第三个稀缺之处，是对优秀人才的极致发掘：不设门槛，公司一把手在内的顶级评审团直接对话创客。
据了解，此次AI黑客松“报名即审”，不设学历、背景与专业门槛，只看创新能力与动手实践能力，招募范围覆盖了180多所高校和40多个开发者社群。
从安克创新创始人阳萌、极客邦科技创始人兼CEO霍太稳、WaytoAGI创始人AJ、线性资本创始人兼CEO王淮，到亚马逊云科技中国区行业集群总经理田锋，从智能硬件巨头、云巨头到AI开发社群、投资机构，AI黑客松的创客们能够获得与行业顶尖评审阵容直面对话的机会。
第四，是优秀的参赛者和作品都“有去处”。
诚然，超过18万元的总奖池、各赛道冠军3万元的奖金，对于创客们来说都有着十足的激励性。优秀选手有机会直通终面，并获得SSP Offer保留机会；入围选手也将优先进入安克高潜人才池。
在丰厚奖池和成长激励之外，安克AI黑客松更核心的吸引力在于优秀作品“有机会进一步进入真实产品线”，或对接合作资源。
相比大部分模拟赛事纸上谈兵式的“脑洞秀”，行业真正需要的是能把“空中楼阁般的AI创意”转化为“物理世界可跑通产品”的真实土壤。安克这场黑客松让创意从阶段性成果走向更长期的落地探索，拥有长期的延展价值。
正如亚马逊云科技汽车与制造行业架构师总监黄鹏飞在开幕演讲中所说，安克的基因就是“把一个品类做到极致，再把能力扩展到下一个品类”，这种“较真”的精神是贯穿始终的。
▲亚马逊云科技汽车与制造行业架构师总监黄鹏飞
结语：AI时代的的下半场，答案在真实场景里
纵观行业发展，这场黑客松更像是当前AI行业演进的一个“缩影”：大模型狂飙突进之下，一味卷参数、卷跑分、卷排名已经成为过去式，行业核心关注焦点逐渐转向：AI能否在真实世界中落地应用、实现商业价值。
从一颗别在领口的录音豆、一台驻留快递站点的智能摄像头到一套在欧洲家庭中高效运转的智能储能系统，24小时集中开发、精彩路演后，这些优秀作品真的有机会进入真实产品研发中。真实的硬件正成为AI的“感官”和“四肢”，AI也不再是贴在硬件上的一项功能，而是从一个具体场景里长出来的解法。
对个人Builder而言，这意味着能力坐标正在移动。“会写代码”已不是拉开差距的地方。从这次的冠军作品里能看到一条共同路径：从身边具体的问题出发，把方案放到真实的数据和设备上验证，并尽量复用已有的能力，而非从零重造。当写代码的时间被AI压缩，写得快不再是优势，判断得准才是。
对于硬件行业来说，这或许也是AI时代做硬件的一种答案。当模型可以租用、代码可以生成，硬件的价值不再只取决于参数和品类，而是越来越取决于它能把AI带进多少真实场景、解决多少真实问题。设备在哪里，场景就在哪里，AI带来的价值提升也就在哪里。
一场优秀的黑客松，比赛结束绝非终点，而是无数创意腾飞的起点。黑客松不是一场PR式的营销噱头，而是检验一家科技公司AI技术底座硬度的“试金石”。安克的这次尝试未必是唯一答案，但它至少说明：AI时代的创新，终究要在真实场景里接受检验。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
Intellectual (WeChat Official Account:zhidxcom)
Author, Zheng Yuen Peng
Edit Indigo
Now, imagine something like this:
When you encounter a place that you don't understand, two taps of taped beans in the neck, after class AI will start from the moment you click, give you a careful answer, without raising your hand, without interrupting anyone;
When the delivery is lost, the AI system directly uses camera-by-cam analysis, intelligently locks out suspicious persons, rapidly locates, and produces a fully integrated video, saves considerable manual review time and significantly reduces communication costs;
In European households across the ocean, the system optimizes the optical economy through AI, reduces solar waste and saves users hundreds of euros a year for electricity;
Today, these scenes are by no means empty talk, but are the best work by the contestants in the 24-hour Ankai Hacksson competition. These works have mature programme ideas, even based on real Ank smart hardware products, which actually “run up” on the scene of the competition.
Behind the scenes and works is a big match: more than 1,800 applications from more than 180 higher education institutions and developers' communities were selected from the floors and, ultimately, 33 teams were killed in the bottom-line finals, and the transition from ideas to demonstrable and verifiable results was completed within 24 hours, with four champions winning.
In the past, making a 24-hour run-through version of this is itself the biggest sight of hackers. But today, with the aid of AI, the cost of writing a Demo quickly is close to zero. The Stack OverFlow 2025 survey showed that 84 per cent of developers were using or planning to use AI tools. The real scarcity of hackers is no longer code capability, but real problems worth solving, access to the real world, and where to go when the work is done.
Ank's first hacker-concert offers a new, visible sample.
I. The value of 24 hours is being redefined
“The Internet changes distance, the AI era changes time.” This is Anke Innovation CIO Neil's judgment of AI.
That's what Ank did first. Anke Innovation 2026 semi-annual reports show that the total token consumption of the company over the past year was 30 trillion, and the daily average of employees was 450 billion, a 30-fold increase over the same period; AI assumed over 95 per cent of the base code generation, saving about 80 per cent of the development hours.
In the workshop sharing at the Harlem Pine Contest, we see that the creators can perform on the Launch platform on a fast-track line, which can be directly transformed into accessible links, from a 360-degree hardware decomposition demonstration, a simulation of product use, an interactive demonstration of pixels, to the generation of a real-time updated BI view board for the docking dynamic data source, with all the functionality of the platform.
This excellent product originates from the high-speed iterative development of the “AI infrastructure” within Ank: Auxiliary development based on AI, the first version of the prototype was completed by the team in just a few days in 25 years; during this year's clean-up holiday, the team completed full-functional go-live only three days to be officially rolled out throughout the company. From a few days to three days, it was Neil who was compressed by the word “time”.
It's in the 24-hour hacker-concert system, which means that when the time for writing codes is compressed by AI, the difference between the players is no longer who knocks faster, but something outside the code — creativity is not feasible, choice is made using the scene, user pains and problems are not captured.
As LLM has soared, the focus of industry ' s core attention has shifted to the ability of AI to apply and achieve business values in the real world.
The real scenes, real problems, real products of SDK were opened to the contestants by Ankhulsson, which allowed many of the best ideas and works to produce real growth soil, behind which was the long-term commitment of Ank to AI and the active quest for innovation.
So, what does the “real question” and “real interface” really open the gap outside the code look like? Let's focus on this hacker pine champion.
How can the ideals be translated into reality?
It depends on how these champions win, first they change the ruler. In the view of AJ, founder of the WaytoAGI, the model determines the ceiling of AI, while the scene determines the value of AI. In other words, what really opens the gap is that everyone now has the ability to use the code, whether the contestant is able to find “the real problem that deserves to be solved” and whether he is able to put it in “the real world that can run”.
Founder of WaytoAGI, AJ
This is confirmed by the scoring criteria for this competition. Product innovation accounts for 35 per cent of primary AI and 35 per cent of the value of the user as well as of the scenery, which is nearly 70 per cent. And what finally emerges, and is remembered, is the actual work of the word "real." It can be innovative in addressing real needs and has the value of landing.
In a two-day event, we were on the scene and we saw the creators ferociously exploring, convulsing and fighting all night. In the final stages of the road show, there are many impressive pieces of work coming out of the four main tracks: smart recording, smart security, battery power, smart service.
In smart recording tracks, the Soundcore soybean team has developed a unique AI-assisted classroom learning programme using the Anke Soundcore soybean function and the context capability of understanding, memory and personalization services behind AZLLM.
In practical life learning, team members find that many students are ashamed to raise their hands and ask questions because they are afraid to disturb others, which affects their digestion of classroom content.
To this end, they use AI technology and smart recording devices to achieve targeted AI answers to questions and to improve learning efficiency and the quality of teaching. At the same time, the product provides teachers with a better understanding of the student ' s state of learning, timely feedback and analysis of the effectiveness of teaching, and further complements the optimization of teaching strategies.
From real teaching pains, real recording of bean products to the birth of "breathing bean", with AI smart hardware and AI technology, teams turn ideas into reality.
In particular, the directors of the panel's Anke recording beans team made a presentation at the site, in which they noted that the pains addressed by the work were highly consistent with the functions of their actual research and development. This means that the direction of the work is indeed feasible and that it is expected to be validated in the future.
In the area of intellectual security in the long-term deep-droping of the Anke overseas market, the “Tracking Station” team has focused its attention on the difficult logistics problem of “fast delivery”.
Using an Anker eufy smart camera, open-product SDK, and AI technology, the team achieved efficient identification and tracking of lost couriers through smart analysis of multiple-road video streams, addressing the pain points of poor and inaccurate manual clearance in traditional logistics surveillance.
In particular, the team members noted that, after testing, they had found that Ank ' s camera had a simple, low-cost feature. In practical applications, the dynamic tracking and time keyword extraction functions of the Ank device are most responsive to demand. Using the functionalities already available in the Ank product, the team streamlined the landing process for the entire product and addressed the problem of low-efficiency and multiple-road distribution.
As noted earlier, the Ank big product is used by actual consumers and can yield substantial real feedback. These feedbacks can provide real data soils for the construction of smart services.
The OntoCare team, which is the champion of the Smart Services Track, is able to bring SOP up to date quickly by building service and decision-making features, adding business rules and SKU in the form of spectronodes, and is better able to avoid the issue of bottom code re-engineering in the face of changes in multinational policy regulations than the traditional WorkFlow and Prompt programmes.
In addition, the team has developed a self-evolutionary system, driven by red-blue confrontations and cases, which has increased the problem-solving rate from 83 per cent to 98 per cent.
In the post-game interview, we learned that the project was inspired by the challenges that Ank faced in the actual delivery scene, and that the team found pains in the real customer service system and was tested in real post-sales data.
In a good team case, we can see that, based on AI technology and the underlying support of Ank's smart hardware, SDK, Builder can quickly turn his ideas into reality. More importantly, these ideas stem from real problems in the real scene and have the opportunity to evolve into technologies that millions of users can use.
In today's wave of AI entrepreneurship, the biggest trap is often “with hammers and nails”: first with AI technology, then with a hard set of applications. And the excellent work on Ank Hawthorne is based on the number of Ankbillion devices, applications that, once landed, can form a data wheel in the real scene and further accelerate the antecedents of AI technology, which is the great potential that Ank has shown in the AI era.
iii. Why is it so rare for Ankei Hawthorne to let imagination land and inspire “real” wings?
Why is it so hard for us to see these wonderful creations in the other hacker-concerts? This makes it necessary to mention the “scarce” of Ank AI hackers, which stems from four key aspects:
First, the theme, scene, product, technology and the final evaluation criteria revolve around the word “real”.
The four tracks are based on the real product line and real business needs of Ank. The fields of smart recording, smart security, charging power, smart passenger service have almost one common feature: A track that is by no means short-lived will require a large amount of long-term inputs, solid product layouts, solid market shares and users.
These are not paper issues, but the real products that Ank has been doing for more than a decade, that tens of millions of families around the world are using, that are already on the market, are sourced water and wood.
Today, when AI is strong enough, popular enough to work, the focus of competition has shifted from technology to scenes.
The second scarcity is the willingness to open the “home” to the bottom of the contestants.
Ank dared to open the interface that is serving hundreds of millions of users around the world to the contestants of the hackers, demonstrating the company ' s strong technical confidence.
These are truly useful, user-friendly, practical SDKs for the Builders, and these key capabilities are rapidly mobilized through SDK, allowing many ideas to be realized in real products. This is in addition to the very high level of completeness and security of the ANK bottom technical interface.
The contestants are not starting from scratch, but rather on the floor of mature technologies and products. Throughout the industry, the courage of Ank to open the bottom of the race will undoubtedly make this company more attractive to innovative talent worldwide.
The third area of scarcity is the discovery of excellence: no threshold, and direct dialogue between top assessors, including companies.
It is understood that this time, AI hackers, “on-call” with no educational, background and professional threshold, is based only on the ability to innovate and practice hands-on, with recruitment reaching more than 180 institutions of higher learning and over 40 communities of developers.
From Yang Ying, founder of the Ank Innovations, the founder of Teletech and CEO Hao, AJ, founder of WaytoAGI, founder of linear capital and CEO Wang Huai, to Tianbai, head of the China Industry Cluster in the Amazon Clouds, from the intellectual hardware giant, the cloud giant, to the AI Development Community, the Investment Agency, the creators of Ai hackers were given the opportunity to speak directly to the top of the industry.
Fourth, both the good contestants and the works are “go somewhere”.
To be sure, the total prize pool of over $180,000 and the $30,000 prize for the champions of the different tracks have been very stimulating for the fans. The best contestants have the opportunity to go straight to the end and have the opportunity retained by SSP Offer; entryists will also be given priority access to the Ankgau Pond.
Beyond the rich award pool and growth incentives, the more central attraction of Anke AI hackersson lies in the “opportunities for further access to real product lines” or the matching of collaborative resources.
What the industry really needs, compared to most of the simulator paper-style “brain-hole show”, is a real soil that can transform “air-house-like AI ideas” into “physical world run-through products”. The hacking of Ank allowed creativity to move from a phase-out to a longer-term landing, with long-term incremental value.
As Huang Peng, Director General of the Automobile and Manufacturing Industry of Amazon Clouds, put it in his opening speech, Ank's genes are the spirit of “maximizing one species and extending its capabilities to the next”.
Wong Peng Fei, Director General of the Automobile and Manufacturing Industry, Amazonic Technologies
Conclusion: The second half of the AI era, the answer is in the real scene.
Throughout industry development, this hacker pine is more like a “microform” of the current evolution of the AI industry: the LLM surged, and the past has become the norm, with the industry’s core focus shifting to whether AI can be used in the real world to achieve business values.
From a taped bean with no access, a smart camera with a delivery site, to a system of intelligent energy that works efficiently in European households, these excellent works have a real chance of getting into real product development after 24 hours of intensive development and showcasing. Real hardware is becoming AI's “sension” and “tangular limbs”, and AI is no longer a feature attached to hardware, but a solution from a specific scene.
For the individual Builder, this means that the capability coordinates are moving. The code of writing is no longer a place to open a gap. From this championship, we can see a common path: From the point of view of specific problems around them, programmes are validated on real data and equipment, and existing capacities are used as much as possible, rather than from zero. When it's time for the code to be compressed by AI, writing quickly is no longer an advantage, it's the right judgment.
For the hardware industry, this may also be an answer to the AI era for hardware. When models can be leased, codes can be generated, the value of hardware will no longer depend solely on parameters and class, but increasingly on how much it can bring to AI and how much real problems it can solve. Where the equipment is, where the scene is, where the value added by AI is.
An excellent hacker pine, the end of the game is by no means the end, but the starting point for countless ideas. Hackers pine is not a PR-style marketing boom, but a “test of gold” to test the hardness of the technical base of AI, a technology company. Ank's attempt is not necessarily the only answer, but it suggests at least that the innovations of the AI era are ultimately tested in the real scene.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.