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 134

A Tiger Behind a Smile

1,651 words 6 min read 2 views 1mo ago Source ​
Skip to the end of the chapter
Prev
Next

When the screen lights up, the background on the other side is a large floor-to-ceiling window, through which you can see the outline of the distant mountain ridges, which are burned into a warm orange by the setting sun.

Ryan Miller looks younger than in the photo, with blond hair, blue eyes, and a high nose. Dressed in a well-fitting shirt and suit, he looks neither like a tech geek nor a business leader, but rather like a movie star.

Han Luyi opened his Sight.

[Ryan Miller, 41 years old | CEO of NeusAI]

It's surprising that information could be detected across the Pacific Ocean.

There is little information about the Sight, no emotion or hidden emotion whatsoever, making further interpretation impossible.

If we continue, we might find some factual information, but since we can't read minds, there's no rush.

(The following dialogue is in English)

"Ni Hao," Ryan smiled at the camera, looking warm and cheerful, and greeted us in oddly accented Chinese before quickly switching back to English, "Finally we meet, Louis. I showed that video to the whole team, and everyone was shocked." "Thank you," Han Luyi said, "We didn't expect the video to have such a widespread impact."

"Some people on X call you a mysterious Eastern power," Ryan said with a laugh, "I don't think that's an exaggeration."

Han Luyi smiled politely.

The small talk ended quickly.

Let me show you something first.

Ryan shared his screen, and a video appeared on the screen.

The video shows an IDE interface.

The user typed the following message in the dialog box. "Add One user permission system to this project."

Mercury begins its work.

It first scanned the entire project's directory structure and existing code, and then automatically planned the implementation steps: create data tables, write authentication middleware, modify existing API interfaces, and add a front-end login page.

Every step is executed automatically. The code is written directly into the file, runs tests automatically after completion, and if a test fails, it is modified and run again. There is no human intervention throughout the entire process.

From input to passing all tests, three minutes.

This is a prototype of the automated workflow that Neus AI is developing.

Han Luyi very seriously.

This thing's coding capabilities are truly impressive. When adding features to an existing project, it can automatically understand the context, plan the steps, write code, and run tests—the whole process is very smooth.

Furthermore, it leverages its robust coding capabilities, using tests to ensure code quality and bypassing reliance on code detection engines like BugKiller. However, there is a problem.

This is purely a developer tool. Users need to have an existing project, know what the "user permission system" means, and be able to understand whether the code it generates is correct.

In other words, this is entirely a tool made for developers, just like BugKiller was back then.

Open Things does something different. Open Things deals with people who don't even have the concept of a "project." They start from scratch and turn a vague requirement into something usable.

The two routes address different problems, but if combined...

After the video finished playing, Ryan switched back to his own face.

"This is still in the demonstration stage, far from productization," Ryan said. "But we are very certain about the direction: in the future, it won't be humans operating AI, but humans supervising AI."

Han Luyi asked, "Is this demonstration using your own model?"

"Yes, Neus Mercury, our general model." Ryan finished speaking, then changed his tone, "But to be honest, the model is only half the battle. Task breakdown, tool scheduling, self-verification—the workload of these scheduling layers is even greater than that of the execution phase." Han Luyi noticed a term: scheduling layer. Ryan had said it casually; in software engineering, it has a specific meaning, but in the field of AI, it might have a different connotation.

Neus is not just building models; they are building a complete execution framework.

"Getting back to the main topic," Ryan sat up a little straighter, "What I want to talk about today is cooperation."

A slide appeared on the screen with the title "Neus X Source Code: Global Code Generation".

"Open Things's product loop in code generation scenarios is one of the best I've ever seen," Ryan said. "The UI, user experience, and task completion rate are all better than what we could do ourselves."

This is just flattery.

"But your backend model has limitations in English-language scenarios." He paused, "You should know this better than I do." This was another blow.

Ryan continued, "Our proposal is: to jointly develop an overseas version of Open Things. We'll replace Heavenly Craft with Neus's universal model as the backend engine for the overseas version of Open Things. Open Things will maintain its product advantages, while Neus will provide a stronger model base."

This is another treat.

"You're quite skilled at the push-pull technique," Han Luyi himself.

The slide flipped to the next page, which showed the framework of the business model.

"Neus will cover the model costs, your team will handle productization and localization, and revenue will be split proportionally—we can discuss the specifics." Ryan looked at the camera, his tone calm but serious. "With your product loop and our model, we can launch within a month." One month.

Han Lu quickly processed these words in his mind.

On the surface, this is a very tempting plan: Neus provides the model and funding, while Neus provides the product and user understanding through the Source Code, allowing each party to leverage their strengths and quickly expand overseas.

But he noticed Ryan's choice of word, "flat".

Ryan's ambition goes beyond simply creating an overseas version of Open Things's code backend; he wants to transform Open Things into a vertical application on the Neus universal platform. This universal platform might be running code generation today, project management tomorrow, and customer service, operations, and finance the day after. For him, Open Things is the first domino to fall.

What can you gain from the Source Code? Money is one aspect, but it's probably not the most important one.

Being able to access world-class models for free and obtain interaction data from overseas users to refine the product experience is extremely important for what Han Luyi wants to do next.

Most importantly, the largest cost—the inference cost—was borne by Neus.

After thinking it over, Han Luyi asked his first question, "How do we implement a data isolation solution? Can we obtain user behavior data from overseas users to iterate on our product?"

Ryan was clearly prepared for this. "Compliance requirements dictate that data processing must be done locally, but we can fully share the anonymized usage statistics for product optimization."

Su Niannian quietly took notes in her notebook, occasionally glancing up at the screen.

She didn't interrupt, but Han Luyi knew what she was writing down, including what Ryan said and what Ryan didn't say.

The conversation entered its fortieth minute.

The atmosphere was great. Ryan was very articulate, logical, and relaxed yet not frivolous.

Han Luyi initially understood why Neus was able to secure over 500 million in Series B funding; besides his impressive resume, he also possessed charisma and approachability. Then Ryan abruptly changed the subject.

"I have one last question." His tone wasn't as relaxed as before; he leaned forward slightly, his eyes sharp—but because he was looking at a screen, the camera captured something rather strange.

"Your product is one of the best at the application layer. We conducted a comparative test with two hundred United States users, directly comparing Open Things and our prototype, and Open Things's performance was significantly better than ours."

Han Luyi was waiting for his turning point.

"But I'd like to know, what's your next direction?" Ryan's voice wasn't loud, but every word was clear. "Source Code, and you're going to make models too?"

"I'm not asking this as a test," Ryan said, "But a very real question. We can Data Sharing and technology, but not with our competitors. If we're going to collaborate, the contract will include a clause stating that the Source Code cannot use the data and technology obtained through the collaboration to develop or train its own Large Model."

Han Luyi did not answer immediately.

Ryan was asking about a very reasonable business term: a data firewall between partners, something that wouldn't be unusual at any negotiating table. But Han Luyi heard more than just that.

The model he's secretly working on, the Universal Intent Understanding Base, is precisely the universal version of the capability Ryan just praised. Why is Open Things stronger than Neus in intent understanding within code generation scenarios? Because of product refinement, the accumulation of user data, and the deep cultivation of the Heavenly Craft model in Chinese scenarios.

However, if we extract this capability from the code scenario and generalize it to all scenarios...

That's not just a product-level advantage; it's a revolution in the user interface.

The automated workflow that Ryan just demonstrated is missing this very element.

If Neus also understands this, they will inevitably compete head-on.

We've considered the benefits of cooperation, but what about the risks?

On the surface, the two companies each contribute their strengths and Data Sharing.

However, Neus obtained this data, allowing him to train his own model and gain experience in developing such applications. The Source Code, cannot use this data to train a model.

When the time is right, Neus can kick out the Source Code at any time and write his own Open Things.

Although there was no indication from the Sight, this Ryan is no fool with too much money.

He's a wolf in sheep's clothing.

"Do we need to create a model for the Source Code as well?"

Su Niannian stopped writing when Ryan asked the question.

She didn't look at Han Luyi, but her fingers gently tightened around the pen cap.

Ryan was still waiting for Han Luyi's reply.

End of Chapter
Prev
NovelSummaries
Next

Explore the wiki

Han Luyi
Su Niannian

Comments

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