Luo Yonghao tries TRAE in Double 11 livestream: you can build software in minutes without knowing how to code
Luo Yonghao demonstrated the full-chain AI development platform TRAE in his Double 11 livestream, claiming he cannot write a single line of code, yet in minutes generated a short-drama research report with charts, turned more than 50,000 characters of material into two sets of PPTs, and built a working voice input method app from a hand-drawn sketch. TRAE's first-month trial price is 9.9 yuan; as of June this year, it has more than 10 million registered users worldwide and 1.6 million monthly active users.
Topics
Body · AI translation
Talk about living long enough to see everything—the most tech-heavy thing about this year's Double 11 turned out not to be figuring out how to stack discounts.
First, the top livestreamer was enthusiastically hawking two trillion Tokens, and right after that I scrolled into Lao Luo's livestream—
The full-chain AI development platform TRAE, which Luo Yonghao calls "Cui Laosan," rolled out a first-month trial price of a sweet 9.9 yuan.
The fifty-something old man who claims he "can't write a single line of code" propped up a laptop and cast his screen on the spot, showing off his moves in front of tens of thousands of viewers:
In a few minutes, generating a short-drama research report complete with data charts;
Reworking over 50,000 words of dense material into two sets of launch-event-grade PPTs;
And from nothing but a rough hand-drawn sketch, cobbling together a genuine, fully functional voice input software.
Unbelievable—AI Coding, which is usually repeatedly benchmarked by programmers in tech forums and nitpicked over performance metrics, was overnight pushed onto the mass-frenzy livestream selling stage.
So? Why are the big livestreamers collectively selling deep, mysterious tech products this year?
In past Double 11s, all we grabbed was toilet paper stockpiled by the case, skincare products whose discounts we couldn't figure out, digital hardware with all sorts of specs...
This year the style is different.
Li Jiaqi's livestream of two trillion Tokens let everyone see for the first time that AI can actually be "stockpiled by the catty."
But lively as it was, many ordinary people, after jumping on the bandwagon and placing orders, instantly had a row of question marks pop over their heads:
I bought a pile of cold, lifeless Tokens—what on earth can I do with them in daily life? Use them to play word-chain games with a large model to kill boredom? Or casually search for material to pad out a document?
The public has grown desensitized to pure technical parameters and abstract numbers far too fast. Once the novelty-seeking curiosity of onlookers passes, the only thing everyone really cares about is:
Can this thing turn into something I can actually use in my hands?
These two livestreams have actually hit a cross-section of the industry.
According to the authoritative firm Gartner's 2026 Generative AI Hype Cycle:
The market has already passed the early phase of inflated concepts and is accelerating toward a pragmatic stage that values real-world results.
Growth in large models that merely chat and answer questions has peaked, while capabilities tightly tied to business scenarios—such as code generation and application building—are drawing more attention.
The industry directly calls 2026 the inaugural year of agents, and even the Cyberspace Administration of China has specifically laid down rules for agents.
Anthropic's coding trends report early this year also mentioned that software development is undergoing its biggest paradigm shift since the invention of the graphical interface:
"Anyone can become a developer" is no longer a slogan on a PPT, and the programmer's role is slowly shifting from "writing code by hand" to "directing AI to do the work."
Beyond selling raw-material businesses like underlying compute power and chat windows, the industry has also begun to output finished-product capability that directly delivers usable tools.
So look again at these two livestreams: what Li Jiaqi sells is raw material, compute quotas; what Lao Luo sells is finished product, tools you can use right away.
What these two livestreams hit is exactly this watershed from selling conversations to selling results.
As a Coding Agent that takes in compute power and directly produces applications, it bridges the gap from raw material to finished product.
Software development used to be seen as an arcane art, but under the whole-network gaze of the livestream, it was compressed into a process as natural as typing and chatting.
Viewers understood too:
What was put on the shelf this time wasn't conversation quota that gets consumed at any moment, but a set of digital productivity that can turn any brainwave into a real tool at any time.
The public's guard against technology also loosened on the spot.
TRAE possesses multi-file awareness, automatic environment configuration, and underlying local code scheduling capabilities.
In the past, the most off-putting, tedious steps in software engineering for laypeople—
English text is being prepared.
To put it more plainly, it can let you build software purely from an idea.
And what kind of person is Lao Luo?
Everyone online knows his nitpicking persona; with him on stage, the whole sales session became a high-pressure test.
As a result, this earnest layman openly stated:
I can't write a single line of code. In the past, making software required recruiting designers, product managers, and a bunch of engineers, and communicating with them was like being a servant; now with Cui Laosan, I can change it a hundred times and it won't even roll its eyes.
In the livestream, Lao Luo cast his screen to show the three major results he had cobbled together in a few hours, each hitting the daily pain points of working people.
Facing the short-drama boom, Lao Luo directly gave TRAE instructions, requiring an analysis of the short-drama market from January to October this year.
AI took less than 4 minutes to spit out a plain-text report. Lao Luo immediately turned into a picky boss and started nitpicking:
Text alone won't do—import it straight into Feishu, use pie charts where pie charts fit, bar charts where bar charts fit; it has to look like I'm seriously bluffing the boss, but I can't actually fool the boss—the data must be solid and reliable!
Lao Luo's computer also unexpectedly froze once midway; after restarting, he told the AI "it just crashed, continue completing the task," and TRAE not only didn't mock Lao Luo with "long live understanding," but steadily picked up the progress and within minutes generated a Feishu document rich with both text and graphics.
Even better were the subsequent workplace-survival combo moves.
Lao Luo had it do a stress test from a management perspective, reverse-engineering a response checklist, laying out predictions in a three-tier structure of "one-sentence core answer + detailed data support + follow-up Q&A."
Now at Monday's regular meeting he'll be fully confident~
Faced with cutting-edge industry material, Lao Luo fed a deep document of over 50,000 characters straight into TRAE to read through and digest, asking it to organize it into a PPT, in two versions: a "speaker version" and a "self-reading version."
Each page of the speaker version kept only the core points, flowcharts, and cases, with all the detailed reasoning stuffed into the speaker notes below; the reading version was complete with text and images and logically self-contained, convenient for others to read silently.
Lao Luo also specifically made requests:
Do it quietly in the background for a while, don't grab the foreground graphical interface and delay me from doing other things; change the format to our company's widescreen 21:9, and use a blue theme.
TRAE obediently took the whole order, and in no time two neatly formatted documents were delivered.
Of course, the most fun part was a tool Lao Luo came up with himself: Mantou Voice.
Because the Microsoft mouse in his hand is jokingly called the Mantou mouse, he wanted to make use of that idle middle scroll wheel button:
In any software, long-press the middle scroll wheel button and a small floating window pops up at the cursor to start speech-to-text; drag and release onto the punctuation or send button, and input is completed in one go.
Lao Luo first hand-drew an extremely sloppy wireframe sketch and shoved it at TRAE; TRAE understood immediately and not only recreated the interaction concept in the sketch, but also proactively broke down the button functions, failure fallback plans, and cross-application coverage.
Not only did it produce a finished product in one go, TRAE can also export the project package for secondary development at any time.
Lao Luo realized he had picked the wrong model at the start, so he falsely claimed he needed to switch computers (which was not actually the case...), and had it package up the source code and project documentation for easy device switching.
Fortunately, TRAE preserved all the project progress, saving this veteran product manager's dignity...
The final working local software can be used in browsers, chat windows, and even Feishu; hold the scroll wheel and speak, and the text appears on screen accurately in real time; casually swipe to the right-side button and release, and commas and question marks are automatically filled in, and you can even directly trigger Enter to send.
I have to say, excellent product philosophy—we're all fans of Brother Long's tech (doge)
After watching the livestream, many people's first reaction is surely just to check out the novelty, and right after that their hearts itch:
I'm not a programmer—can I really use it to do something?
Yes, and the barrier is much lower than you think.
In the past, making a small piece of software meant either self-studying syntax for years and staying up until you went bald, or finding a technical team to outsource to, with starting costs often tens of thousands of yuan.
Now, low-level code writing is deeply managed by TRAE, and the cost for ordinary people to validate an idea keeps dropping; with a few hours of spare time at most, you can mold the idea in your head into a clickable, interactive prototype.
As of June this year, TRAE's global registered users have exceeded 10 million, with 1.6 million monthly active users, and a considerable portion of users open it to work nearly 5 days a week.
So what do ordinary people use it for?
Two high school students who love playing badminton spoke from experience!
Because hiring professional coaches long-term is too expensive and they usually can't see their own movement details clearly, despite knowing nothing about programming, they used TRAE and sensor data from phones, watches, and earphones to build an app that analyzes swing movements and gives guidance—and it has even been officially listed on app stores;
There was also a young man torn about whether to install surveillance cameras at his grandmother's home.
If installed, he could know promptly when the elderly person falls, but at the thought of Grandma being watched by a camera all day, he really couldn't bear it.
In the end, he used TRAE to build the Fanglian Guardian Assistant, which does not rely on camera audio or video recording, but only uses pill boxes and door magnetic sensors to judge the situation, pushing reminders only when medication is missed or there is an abnormal outing, preserving the dignity and privacy of the elderly in their later years.
The global developer community GitHub mentioned in its annual report "State of Octoverse":
AI is lowering the development barrier, and a large number of citizen developers without traditional programming backgrounds are beginning to participate in building applications; over 40% of PRs (pull requests) on the platform are initiated by AI Agents, and open-source contributions from non-professionals are increasing rapidly.
Although TRAE hasn't smashed the rice bowls of senior system architects, nor will it make you rich out of thin air, it has genuinely filled in the development gaps for independent creators who lack technical partners.
What blocks ideas from landing is no longer incomprehensible terminal errors; as long as you keenly capture real pain points around you and express them clearly, you have a chance to land them painlessly.
Administrative and operations staff sitting in offices, tortured daily by spreadsheets in all sorts of bizarre formats, can absolutely use it to whip up a dedicated batch data-cleaning script;
Independent photographers and illustrators no longer need to buy expensive website templates; with a few natural language interactions they can build a portfolio page with their own unique style;
Small e-commerce sellers can casually customize a set of automated accounting mini-plugins tailored to their shop's specific commission rules.
...
One person, with a clear business concept plus the seamless execution of tools, can support a process that used to require a small team to run.
Don't say it—I really spotted a need while watching the livestream, and immediately TRAE'd it!
I built a cliff cabin with my bare hands, ready to drag my "enemy" over to talk through our issues (doge).
What does everyone want to whip up?