Skip to main content
  • Home
  • Library
  • Read
    • Popular
    • Charts
    • Browse
    • Categories
    • Wikis
    • Glossary
  • Request
  • Activity
  • Leaderboard
  • Help
Our Apps
  • DaoSearch
  • Narratorr
Social
  • Discord
  • Patreon
The AI Wave Tycoon's Rise
Chapter 226

What Is Yufeng?

2,328 words 8 min read 1 view 3mo ago Source ​
Skip to the end of the chapter
Prev
Next

Upon hearing Jason's words, the host raised an eyebrow with interest.

This was news he hadn't heard before.

So, NeusAtlas really wasn't the first one?

But can China truly create a groundbreaking product before Silicon Valley's tech elites?

Michael, who was standing next to Jason, felt a chill run down his spine when he heard Jason's question.

It's over, it still happened.

NeusAI had already discussed this internally; their goal in this publicity stunt was to gain a voice in the Agent field and make a name for themselves.

In Silicon Valley, they were indeed the first to launch the service. After the press conference, the closed beta test was gradually opened up, and the topic was generating a lot of discussion.

It can be said that NeusAI has already achieved more than half of its original goals.

The biggest hidden danger now is that some people are bringing up the fact that the China Source Code Technology's Riding the Wind product was released earlier than theirs.

In fact, Ryan had already thought of this problem. On the very night he received the Riding the Wind test account for Atlas, he set a very short deadline of four weeks to announce Atlas.

His assessment at the time was that, given the complexity of the engineering challenge, Source Code Technology would likely conduct more internal testing and iterate through several iterations before releasing a public beta.

As long as we announce Atlas before their public beta, there won't be any problems, since the impact of closed beta is small and fewer people use it.

Unexpectedly, the public beta launch event for Riding the Wind took place only a week after the closed beta, and the public beta version was much more complete than the closed beta version, making it hard to believe that the iteration was completed within a week.

This resulted in Riding the Wind not only having its closed beta test, but also its public beta test and launch event being held before Atlas.

The only saving grace is that, as China startup, Source Code Technology and as a Chinese-language tool, Riding the Wind did not generate much discussion in the English-language tech world.

However, this forced NeusAI to urgently adjust its timeline and hold an earlier press conference to announce Atlas.

Because of this, Atlas was not yet fully developed when the press conference was held, and only a video demonstration could be used. The entire team worked overtime for more than a week to finally produce the beta version a few days ago.

The problem now is that Jason brought up the topic of Riding the Wind in public.

This involves media outlets and bloggers, and the discussions are edited into podcasts afterward.

Jason himself also represents Google.

A bad answer can have very serious consequences.

Michael also noticed that Jason's question was very targeted: he asked if NeusAI knew about a China company that had made the same tool and released it earlier.

This question points directly at the company, not at Michael himself.

This is probably only his first question.

If Michael denies it outright, it's anyone's guess what other offensive he'll launch next.

Michael responded quickly, saying, "This is a very cutting-edge field, and it's normal for excellent teams to be exploring this direction. I believe that Google and OpenAI also have teams exploring this area."

Michael first made a few evasive remarks, then tossed the question back, "I personally don't pay much attention to teams outside the United States, and I've never heard of Yufeng. Jason, have you used this product? Do you think it's better than United States products?"

Michael subtly shifted the focus here, not saying whether Yufeng is better than Atlas, but rather whether it is better than United States products, in order to incite outrage among others.

Don't assume United States aren't sensitive to this. They are exceptionally sensitive to things like being number one in the world.

Sure enough, the host frowned slightly the instant Michael finished speaking.

Jason seemed oblivious to the hidden meaning in Michael's words. In reality, he was a first-generation immigrant from China who had become a United States. citizen. However, as an outstanding engineer from a small town with a background of excelling in exams, he was not actually good at this kind of backstabbing.

Jason continued, "I haven't used Yufeng's products. I only know from community feedback that their focus is on Agent programming, and they're quite accomplished. Michael, you said you haven't heard of Yufeng, but as far as I know, this company collaborated with you on an AI Platform called Kaiwu, and Yufeng's software uses Neus's Mercury Model in its backend."

Jason paused, then asked, "You really haven't heard of it?"

It doesn't make sense that you haven't heard of the company you're partnering with, does it?

If they released theirs first, and you released yours later, and the two companies have a cooperative relationship, isn't it a bit much for you to claim you were the "first"?

Michael understood the subtext, but he also gleaned information from Jason's words that he could use, which actually put him at ease.

"I'm familiar with Kaiwu's product, and I didn't expect their team to have also explored Agent programming. Indeed, we have a cooperation agreement with a China company. If they also use the Mercury Model to create a usable Agent Product, it proves that Mercury does have merit in this area."

Michael answered the question with a smile.

Since you've already investigated so thoroughly, I won't discuss the order of events. Anyway, the Mercury Model is used in the later parts, so what's there to say?

Mercury is awesome, Gemini (the constellation of Google's Large Model) is just sitting on the side of the road.

Jason was taken aback by Michael's reaction, which was quite different from what he had expected.

At this point, the host picked up the conversation. "This is an excellent topic. Our guests are all professionals in the field of AI. In your opinion, what is the most crucial capability for programming Agent? Is it modeling ability or product engineering ability?"

Jason thought he had dropped a bombshell, but it didn't achieve the desired result, so he was a little unwilling. However, he had already changed the subject. The other two people present were from the marketing field, while he was the only one with a technical background. This question was actually meant for him.

Jason gathered his thoughts and said, "Model capabilities are certainly important, but in software engineering scenarios, models are only one part. The real challenges lie in context management, task breakdown, execution verification, permission boundaries, and how to recover after failure."

This is practical information, and it reflects his own genuine experience in developing Agent applications.

Gemini's capabilities are undoubtedly top-notch among Large Model, but building usable Agent tools on top of this model remains a huge engineering challenge.

Although Jason is unwilling to admit it, the fact is that both Riding the Wind and Atlas have already surpassed the Google team.

After listening to Jason's answer, the host nodded in agreement and then guided the discussion.

Three hours later, a written preview of the roundtable discussion was available. The discussion covered a wide range of topics and was quite in-depth, exploring the significance of Agent in the development of AI and potential future directions.

However, when choosing a title, the editor opted for the one that would generate the most discussion and grab the most attention.

Was NeusAtlas truly the first programmable Agent tool?

The subtitle is more direct.

<Google's AI tools lead says another tool developed by the China team was earlier and is better.>

The video podcast is still in editing and won't be available for at least a few more days, but the article has already quickly gone viral online.

It was first forwarded to X, then quickly posted to HackerNews, and then several programming communities on Reddit started discussing it.

Initially, most overseas developers reacted as follows:

What is Yufeng?

[Never heard of it.]

[Can China also produce products in this direction?]

Then, very quickly, a helpful porter uploaded the video of the public beta test of Riding the Wind.

Someone translated the reviews from the Chinese community into English.

Someone posted a thread showing Riding the Wind users using it to refactor old projects, fix complex bugs, and generate test cases.

Someone even compiled a timeline.

Closed Beta Test of Riding the Wind.

The open beta of Riding the Wind has begun.

Ne. usAtlas demo released.

Once the timeline is laid out, things become complicated.

One comment was quickly pushed to the top.

[If this timeline is accurate, then Atlas was at least not the first public product.]

Someone below refuted this:

China companies are very good at rapid replication.

Someone immediately posted a screenshot:

[Please note the timeline. Yufeng's public test predates Atlas.]

Later, after an emergency discussion with Ryan, Michael had no choice but to speak out.

"The so-called world's first Agent programming tool was all hype in the media. NeusAI has never publicly promoted it and has no intention of deceiving anyone. We are pleased that our China partners can use Mercury Model to create useful tools, and we hope everyone will pay more attention to Mercury Model. Thank you."

After Jason returned, he was called in for a talk by the Marketing Department, who pointed out the inappropriateness of his remarks.

However, the goal of breaking the narrative of NeusAI's exclusive control over Agent has been achieved.

OpenAI, however, doesn't want to get involved in this war of words; they are accelerating the development of their own Agent tool—Code.

After all, it doesn't matter who is the first; what matters is who is the best.

When people in China first saw this story, it was already the next morning, while overseas public opinion was still developing.

The earliest source to repost it was a tech-focused self-media outlet, whose Chinese headline was even more explosive than the original English version.

Silicon Valley is shocked! The world's first programmable Agent tool actually comes from China?

Soon, intense discussions also began on the Chinese internet.

[Holy crap, haven't seen you in a few days, Riding the Wind's gotten quite the brawn?]

[Didn't someone say just a few days ago that Riding the Wind copied foreign works? What else does Xiaoheizi have to say?]

Ne. us is truly shameless; after copying, they even bought articles claiming they were the first to do so.

[To be precise, the original text didn't say they plagiarized, it only said they weren't the first.]

[Seriously wrong! The timeline is right here: the public beta of Riding the Wind predates the Atlas demo, and the two companies even collaborated. What else could it be but plagiarism?]

[Previously, domestic products were certified abroad; now, foreign products are being sanctioned by domestic timelines.]

While some people argued in the comments section, many more began to discuss the Riding the Wind himself.

Many people who had seen or heard of Riding the Wind before but had never tried it have now applied for public beta accounts.

The registration growth rate, which had already slowed down somewhat, was suddenly stimulated again, and in just a few days, the number of registered accounts for the public beta test has reached nearly two million.

BugKiller's enterprise clients have even contacted Yufeng, hoping that Riding the Wind will launch a formal paid enterprise version as soon as possible.

Jiang Yixin, along with her team, urgently devised a pricing plan, drawing on BugKiller's past experience and comparing it with competing products on the market.

This time, YuShi is facing almost the exact same dilemma as BugKiller back then. As leaders in the field, or even pioneers, they don't have exactly the same products to refer to in terms of pricing strategies, so they can only look to similar fields for reference.

Jiang Yixin and decided to set the initial price a bit higher to see how the paid conversion rate went. If the conversion rate was too low, she would adjust it later.

Jiang Yixin was able to make such a quick decision thanks to the management books she had recently been reading, including Jeff Bezos's...

In Bezos's Invent and Wander, a decision-making principle used at Amazon is mentioned—the one-way door/two-way door principle.

The so-called one-way door/two-way door principle divides the decisions encountered in daily management into two categories.

Two-way decision-making allows for rapid implementation after a decision is made, allowing you to observe the results. If the outcome is positive, changes can be made. This type of decision can be made quickly. For example, in a team meeting, if you want to change from a rotating speaking format to a free discussion format, you can test the effect first and then switch back if it works.

One-way decision-making refers to decisions that, once made, cannot be easily changed, and therefore require careful consideration.

Pricing, especially setting a low price, is actually a typical one-way door decision scenario. Once the price is set too low, users will have an anchor price in their minds, making it extremely difficult to raise the price later, and it will also cause customer resentment.

Instead, they set a relatively high price at the beginning, leaving room for self-developed products to lower the price.

After Jiang Yixin finished the pricing plan, she quickly sent it to Su Niannian for approval, and also copied Han Luyi.

Han Luyi, however, had no time to pay attention to his official career.

Although he was sitting in his office, his energy was now completely focused on Tang Jin's training and the deployment of the Data Center.

The Data Center is basically in place, and the adaptation for domestic card inference is almost complete. The migration of Heavenly Craft's multiplier traffic has not caused any problems.

Jiang Songran also conducted some preliminary drills for the deployment of the general-purpose Large Model, and is just waiting for Tang Jin to complete the training before conducting test drives.

Tang Jin's training has entered its final stage, and according to Zhao Wenyuan's prediction, it will be completed in a few days.

Han Luyi was reading the report and pondering his next move when a burst of golden light suddenly appeared before his eyes.

Sight -

It's been upgraded.

End of Chapter
Prev
NovelSummaries
Next

Explore the wiki

Han Luyi
Su Niannian
Jiang Yixin
Zhao Wenyuan

Comments

No comments yet.
AboutStatsAPIChangelogStatusContactPrivacyTermsDMCA
© 2026 RandomTranslator·v3.26·RSS