Skip to main content
  • Home
  • Library
  • Read
    • Popular
    • Charts
    • Browse
    • Categories
    • Wikis
    • Glossary
  • Request
  • Activity
  • Leaderboard
  • Help
Our Apps
  • DaoSearch
  • Narratorr
Social
  • Discord
  • Patreon

Recaps

The AI Wave Tycoon's Rise
VolumesArcsSummaries

Arc 6Building Glutinous Rice Ball

Han Luyi and Zhao Wenyuan trained Glutinous Rice Ball using precisely annotated data, then sought computing power for a new model company after it outperformed Dingsheng’s Kun Yuan. Meanwhile, Jiang Yixin developed Agent with NeusAI’s tools, and Han gave Zhang Haoran access to the model, which advised him on a bank migration report.

  • Han Luyi decided to pursue a general-purpose model despite He Yunshen’s warning, and began gathering data, a team, and computing resources while considering the legal risks of working on a new company alongside Source Code Technology.
  • Han Luyi used Sight to infer users’ real needs from their behavior, annotated thousands of examples himself, and Zhao Wenyuan trained a prototype that improved its intent-understanding results.
  • Han Luyi rejected Neus AI’s proposed partnership terms restricting Source Code Technology’s model development, while Jiang Yixin argued that the company should retain its data and ability to train its own model.
  • Han Luyi and Zhao Wenyuan showed Dingsheng’s Cheng Yuan and Liu Dahai that Glutinous Rice Ball outperformed Kun Yuan on intent-understanding tests, then proposed that Dingsheng invest GPU usage rights for equity in a new model company.
  • Zheng Xiaobo ordered Cheng Yuan to investigate Source Code Technology’s data and pursue a computing-power partnership that required data sharing, while Han Luyi continued annotating data himself.
  • Jiang Yixin and Qian Xiaole built Agent, a tool that connected Mercury Model, Heavenly Craft, and BugKiller to complete coding tasks, and Han Luyi told Jiang to refine it through internal use and feedback.
  • Han Luyi met Zhang Biao, Ma Xiaofei, and Zhang Haoran, then gave Zhang Haoran access to Glutinous Rice Ball, and the model advised him to recommend a phased migration of the bank’s core systems without openly taking sides.
Chapters 127–151 Vol. 2 Read

Ch. 127Afraid of going slow, not fast (6k)

Han Luyi tried to persuade investor He Yunshen to support a move into general-purpose AI models, but He refused, believing Source Code Technology should focus on its current business. Han left the meeting intending to find another way forward, while He arranged closer oversight of the company.

  • Han Luyi explained how NeusAI’s autonomous workflow approach had made him consider building a general-purpose model, but He Yunshen warned that expanding too soon could undermine Source Code Technology.
  • He Yunshen advised Han to focus on tools and user growth, arguing that a partnership with an independent team would still divert Han’s attention from the company.
  • He Yunshen described three signs that a new direction was ready: sufficient resources, early customers willing to pay, and growing consensus among capable people.
  • Han accepted He Yunshen’s answer without revealing his plans and left, deciding to seek another route after the investor refused to support the move.
  • After Han departed, He Yunshen told his assistant that Hongyuan would lead Source Code’s next round and ordered the portfolio team to visit and ensure Han did not spend recklessly.
Read chapter

Ch. 128Leveraging others' resources (Bonus chapter for alliance leader [A Cast-Iron Pot])

After deciding to secretly pursue a new venture rather than wait for He Yunshen’s approval, Han Luyi realized he needed data and training-grade compute more than funding. Qingyue Cloud had no GPUs available until next year, so he turned his attention to the tech giants, which had ample resources.

  • Han Luyi told Zhao Wenyuan that his talks with He Yunshen were still ongoing.
  • Han Luyi decided to assemble data, a team, and resources for the new venture before asking He Yunshen to invest.
  • Han Luyi concluded that training-grade compute was his most urgent need and called Liang Yu about Qingyue Cloud’s GPUs.
  • Liang Yu said Qingyue Cloud’s GPUs were fully booked and new training capacity would not be available until April next year.
  • Han Luyi ended the call and considered approaching top tech giants, which were not short on compute.
Read chapter

Ch. 129The Real Use of a Cheat Ability

Han Luyi analyzed how Reality code Sight worked and realized its value for model training: it could reveal the intent hidden behind users’ surface instructions. After landing, he planned to collect intent-discrepancy examples, fine-tune an open-source model, and use the results as leverage in talks with Dingsheng.

  • At Beijing Daxing Airport, Han Luyi used Sight to see that Chen Jiujiu’s request for a Q4 performance report actually concealed a proposal to establish her data-auditing business as an independent service line.
  • On the flight, Han Luyi reviewed Sight’s limits and concluded that it organized existing information rather than creating knowledge or predicting the future.
  • Han Luyi recognized that BugKiller succeeded because he used Sight to identify vulnerabilities and turned those observations into detection rules that the product could replicate.
  • Han Luyi connected that approach to AI models and decided Sight could reveal the true intent behind users’ prompts, enabling him to create better training data.
  • After the plane landed, Han Luyi messaged Zhao Wenyuan to discuss the plan and told Zhang Biao to return straight to the company.
Read chapter

Ch. 130Consider It a Loan from Me

Han Luyi showed Zhao Wenyuan precise user-intent annotations and planned to use them to fine-tune a model prototype for investors. After failing to secure funding from Boss He, Su Niannian offered the new company a personal loan, while Han Luyi kept another funding plan to himself.

  • Han Luyi showed Zhao Wenyuan a spreadsheet of 287 annotations that inferred users’ real work needs from their behavior in Open Things.
  • Zhao Wenyuan estimated that five to eight thousand examples could demonstrate improved intent understanding in a fine-tuned model, and Han Luyi said he would present the prototype to investors.
  • Su Niannian brought Han Luyi lunch, confirmed she knew his plans, and heard that the prototype would be ready within two weeks.
  • Han Luyi told Su Niannian that negotiations with Boss He had failed, so she offered to lend the new company her remaining sixteen million without taking equity.
  • Han Luyi said he had another way to handle funding but did not explain it, and Su Niannian reminded him about the video conference with Ryan scheduled for the following Wednesday.
Read chapter

Ch. 131Ding Pay

Han Luyi planned to use Sight to create richer annotations from open-source data while Zhao Wenyuan built a training pipeline. Gu Siyue explained the legal and investor risks of Han Luyi and Zhao Wenyuan working on a separate model company, leaving Han Luyi to consider how much trust he could risk.

  • Han Luyi told Zhao Wenyuan he would handle high-quality data annotation with Sight, and Zhao returned to work on the training pipeline.
  • Han Luyi reasoned that Sight could reveal overlooked context in raw data and pair it with more accurate intent annotations than ordinary annotation methods.
  • Han Luyi met Gu Siyue at Siheng Law Firm to ask whether he and Zhao Wenyuan could work for both Source Code and a new model company.
  • Gu Siyue explained that Zhao's unvested three-percent options and He Yunshen's board oversight could make Zhao's outside work risky, while Han Luyi's full-time commitment clause created a potential breach.
  • Gu Siyue recounted how Lü Yun's transfer of Ding Pay damaged investor trust and warned Han Luyi that separating assets from Source Code carried serious risks.
Read chapter

Ch. 132Thanking a Client

Han Luyi warned Zhao Wenyuan that helping with the new venture could cost him sixty million in options, then gave him annotated data to test a prototype without using company resources. As Zhao began training the model, Han discovered that Sight had gained experience and reached 1% progress toward another level.

  • Gu Siyue thanked Han Luyi for his legal consultation by treating him to McDonald's, where they ate together awkwardly.
  • Han Luyi told Zhao Wenyuan that the board might force him to resign and void his options, but Zhao chose to continue helping.
  • Han Luyi and Zhao agreed to build a small prototype without registering a company or using Source Code Technology's resources.
  • Three days later, Han Luyi gave Zhao 8,100 annotated open-source entries, and Zhao realized Han had annotated them all himself.
  • Zhao copied the data and began training the model on his personal computer while Han checked Sight and saw its experience bar had reached 1%.
Read chapter

Ch. 133People Are Gullible and Rich—Come Quickly

Han Luyi investigated why his Sight experience had started increasing and prepared with Su Niannian for a meeting with Neus AI. After testing Neus Mercury, he recognized that its powerful code generation lacked Open Things’ easy-to-use product experience, and the story left him joining a video meeting with Neus.

  • Han Luyi found that his Sight experience bar remained at Lv.3, 1%, and wondered whether its recent increase came from annotations or training.
  • Han Luyi met Su Niannian at the office, where she reviewed Neus AI’s funding, leadership, and business with him.
  • Han Luyi used Sight to inspect Neus Mercury’s code generation and found a few bugs but judged its output high-quality and difficult for ordinary users to deploy.
  • Han Luyi and Su Niannian agreed to hear Neus’s proposal without promising anything and to learn what benefits Neus could offer.
  • At 7:58, Han Luyi clicked the meeting link to meet with Neus.
Read chapter

Ch. 134A Tiger Behind a Smile

Han Luyi considered Neus’s proposal to partner on an overseas version of Open Things, powered by Neus’s model, and weighed its benefits against the risks of sharing data and technology. Ryan asked whether Source Code planned to build models, leaving Han Luyi and Su Niannian to face the question.

  • Ryan showed Han Luyi Neus’s automated coding prototype and explained its model and orchestration framework.
  • Ryan proposed that Neus provide its model and cover costs while Source Code developed and localized an overseas version of Open Things, with revenue shared between them.
  • Han Luyi asked how overseas user data would be isolated and whether Source Code could access anonymized usage statistics.
  • Ryan described favorable results from a comparison of Open Things and Neus’s prototype, then asked whether Source Code planned to build models and proposed a clause barring Source Code from using partnership data or technology to train them.
  • Han Luyi recognized that the partnership could help Neus learn from Source Code while restricting Source Code’s own model development, and Ryan waited for his answer as Su Niannian stopped writing.
Read chapter

Ch. 135Your take is wrong-hear mine out

Han Luyi argued to Ryan that Source Code Technology would focus on applications rather than foundation models, persuading Ryan to propose a partnership while concealing the company’s own model plans. After the call, Han Luyi admitted to Su Niannian that he had been lying.

  • Han Luyi told Ryan that Source Code Technology’s advantage lay in its products and user experience, not in owning the best model.
  • Han Luyi used the history of the dot-com bubble to argue that value shifted from internet infrastructure to applications, while Ryan challenged his comparison and defended models’ data flywheels.
  • Ryan said Google and Meta were pursuing similar automated workflows and proposed that Source Code Technology launch an overseas version of Open Things and jointly build a workflow product.
  • Ryan warned that building a general-purpose model independently would require at least a billion dollars and a year, and Han Luyi said he would consult the founding team.
  • After the call ended, Su Niannian asked whether Han Luyi truly believed his argument, and he admitted he had lied to Ryan.
Read chapter

Ch. 136Far beyond expectations

Zhao Wenyuan reported that the first training results for PX-7B far exceeded expectations in intent understanding, easing Han Luyi and Su Niannian’s concerns about NeusAI’s prospects. Han Luyi decided to seek computing resources from Dingsheng in exchange for equity and reached out to Cheng Yuan to discuss cooperation.

  • Zhao Wenyuan showed Han Luyi and Su Niannian evaluation results that ranked PX-7B first in intent understanding, ahead of GPT and Gemini.
  • Zhao Wenyuan estimated that a hundred thousand high-quality annotations could push larger models’ intent understanding beyond the industry ceiling, and said he would take unpaid leave to finish tuning the prototype in a week.
  • Han Luyi proposed getting computing power from Dingsheng in exchange for equity, and Su Niannian agreed as long as the company retained control.
  • Zhao Wenyuan accepted Han Luyi’s insistence that the model be trained in China and raised no objection to working with Dingsheng.
  • Han Luyi messaged Cheng Yuan to ask whether they could discuss a new project.
Read chapter

Ch. 137Making a name overseas

Han Luyi arranged a meeting with Cheng Yuan to discuss a large-model partnership, while Jiang Yixin learned that Open Things was drawing strong overseas interest despite its lack of international support. She discovered a crude English-language competitor already serving paying customers and went to report it to Han Luyi, who invited her to lunch with Su Niannian.

  • Han Luyi arranged to meet Cheng Yuan the following Tuesday to discuss a large-model partnership, and decided to keep improving Zhao Wenyuan's prototype before the meeting.
  • Jiang Yixin arrived at Source Code Technology and learned from Little Niu that Zhao Wenyuan was taking a week off for personal reasons.
  • Li Ting showed Jiang Yixin an overseas public-opinion report documenting strong interest in Open Things, including hundreds of thousands of views on Li Ting's English tutorial and users tolerating VPNs and high latency.
  • Jiang Yixin read about AppFlow, an English-language app-building service with paying users and international support, and angrily called it a copy of Open Things.
  • Jiang Yixin went to tell Han Luyi about AppFlow, but Han Luyi invited her to join him and Su Niannian for lunch so she could explain it there.
Read chapter

Ch. 138You Silly Girl, You're Moving Up

Jiang Yixin advised Han Luyi and Su Niannian to pursue Open Things’ overseas market while rejecting Neus AI’s restrictive partnership terms. Han Luyi brought her into the upcoming coordination meetings, and Cheng Yuan submitted a partnership report to Zheng Xiaobo.

  • Jiang Yixin told Han Luyi and Su Niannian that Open Things should enter the overseas market rather than waste effort fighting copycats.
  • Su Niannian explained Neus AI’s proposal to provide its Mercury model for the overseas version in exchange for revenue sharing and a ban on training Source Code Technology’s own model.
  • Jiang Yixin warned that the deal could lock the company in, and proposed a short-term partnership only if Source Code Technology retained its core data and could train its own model.
  • Han Luyi asked Jiang Yixin to write up her analysis and said she would join the partnership coordination meetings; Su Niannian praised her afterward.
  • Li Ting told Jiang Yixin she was moving up and should prepare to treat everyone.
  • Cheng Yuan submitted a draft report on the partnership’s direction to Dingsheng CEO Zheng Xiaobo.
Read chapter

Ch. 139A Show Is Coming

Cheng Yuan reported on Source Code Technology’s interest in large-model cooperation, and Zheng Xiaobo decided to use a technical demonstration to discourage Han Luyi from building a model while keeping acquisition open. The meeting was set for the following Tuesday, with Dr. Liu Dahai scheduled to present Prosperity Group’s foundation model.

  • Cheng Yuan submitted a report recommending an initial meeting with Source Code Technology after Han Luyi expressed interest in large-model cooperation.
  • Zheng Xiaobo summoned Cheng Yuan and asked for his assessment of Han Luyi and Source Code Technology.
  • Zheng Xiaobo said Prosperity Group valued Source Code Technology’s application-layer products and user data, and preferred to bring Han Luyi into the company.
  • Cheng Yuan told Zheng Xiaobo that Han Luyi might want to build a large model, and Zheng Xiaobo dismissed the idea as unrealistic for a small startup.
  • Zheng Xiaobo ordered Cheng Yuan to arrange for Dr. Liu Dahai to demonstrate Prosperity Group’s foundation model at the meeting.
  • Cheng Yuan’s secretary confirmed Dr. Liu Dahai’s participation, and Cheng Yuan anticipated a show at the meeting next Tuesday.
Read chapter

Ch. 140The person I like likes me too

Han Luyi spent the day shopping for staff gifts with Su Niannian, whose gestures and Christmas wish hinted that she liked someone. He left the chapter wondering whether her wish was about him after Sight’s experience unexpectedly rose to 8%.

  • Han Luyi saw that Sight’s experience remained at 7% and agreed to meet Su Niannian at Taikoo Li to choose staff gifts.
  • Han Luyi and Su Niannian browsed shops, and he took an instant photo of her that he slipped into his pocket.
  • Su Niannian bought Han Luyi a scarf and wrapped it around his neck.
  • At lunch, Su Niannian said that people sometimes refused to trust an obvious answer, then asked whether wishes like hers came true.
  • Han Luyi glimpsed Su Niannian’s wish, “The person I like likes me too,” and later noticed Sight’s experience had reached 8%.
Read chapter

Ch. 141Among the nation's best

Han Luyi learned that Zhao Wenyuan’s latest fine-tuning had significantly improved Source Code’s model, then decided to buy prizes for the company’s employees. At Dingsheng, Cheng Yuan and Liu Dahai prepared to demonstrate Kun Yuan’s benchmark results to Han Luyi, confident in their model’s standing.

  • Zhao Wenyuan told Han Luyi that Source Code’s model had improved across key benchmarks, and Han ordered him to deploy it for real-world testing instead of further tuning.
  • Han Luyi realized that Zhao Wenyuan’s fine-tuning had raised Sight’s experience gain to 8%, reinforcing his suspicion that model training affected it.
  • Han Luyi and Su Niannian visited a DJI/Hasselblad store, where Han decided to buy three Pocket cameras and three Action cameras for a company prize draw.
  • At Dingsheng, Cheng Yuan met with Liu Dahai, who showed him Kun Yuan’s results and said the model ranked among the top domestic models, close to GPT-4 on intent comprehension.
  • Cheng Yuan instructed Liu Dahai to give Source Code a brief benchmark demonstration, and he waited confidently for Han Luyi’s arrival.
Read chapter

Ch. 142The show begins

Han Luyi and Zhao Wenyuan visited Prosperity Group to discuss a new venture, where Cheng Yuan and Liu Dahai demonstrated the Kun Yuan model. Han then showed them his small Glutinous Rice Ball model, whose interpretation of a test prompt left Liu Dahai stunned.

  • Han Luyi and Zhao Wenyuan met Cheng Yuan and Liu Dahai at Dingsheng Tower for a model demonstration.
  • Han Luyi used Sight to see that Cheng Yuan was confident but suspicious, while Liu Dahai cared mainly about technical research.
  • Liu Dahai demonstrated Kun Yuan, which generated a polished report from spreadsheet data.
  • Han Luyi presented Glutinous Rice Ball, a fine-tuned model with seven billion parameters and fewer than ten thousand training examples.
  • Liu Dahai tested both models with the same work-history prompt and saw Glutinous Rice Ball produce a persuasive weekly report rather than a bare activity log.
  • Liu Dahai opened Glutinous Rice Ball's thinking process and read its inference that the user needed a self-protective report amid possible performance pressure, leaving him alarmed.
Read chapter

Ch. 143One Triumph After Another (Bonus chapter for alliance leader [Lone Step Roars to Heaven])

Han Luyi and Zhao Wenyuan stunned Dingsheng with Glutinous Rice Ball’s superior intent-understanding results and proposed a new company in exchange for computing resources. As they left, Liu Dahai began analyzing the model’s advantage and was summoned to meet Zheng Xiaobo the next morning.

  • Liu Dahai tested Glutinous Rice Ball with a private benchmark question, and it identified the hidden intent and trap that Kun Yuan had previously missed.
  • Han Luyi revealed Glutinous Rice Ball’s intent-understanding score of 85.2, seven points above Kun Yuan, leaving Liu Dahai and Cheng Yuan stunned.
  • Han Luyi rejected Cheng Yuan’s attempt to discuss cooperation with Source Code and proposed that Dingsheng invest GPU usage rights for equity in a new foundation-model company.
  • Han Luyi and Zhao Wenyuan left Dingsheng after asking Cheng Yuan to review the proposal, and Zhao told Zhang Biao that their model had scored 85.2 against Dingsheng’s 78.
  • Liu Dahai began writing a technical analysis of Glutinous Rice Ball’s performance, then Cheng Yuan told him Zheng Xiaobo wanted to meet them the next morning.
Read chapter

Ch. 144The Only Explanation

Cheng Yuan and Liu Dahai reported to Zheng Xiaobo that Han Luyi and Zhao Wenyuan had offered to exchange Dingsheng’s GPU usage rights for equity in a new foundation-model company, after their model decisively outperformed Kun Yuan. Liu Dahai judged the results genuine and attributed the advantage to exceptional labeled data, and Zheng Xiaobo identified ScaleAI as a possible explanation.

  • Cheng Yuan admitted he had not controlled the meeting and described the proposal to invest GPU usage rights in a new company being prepared by Han Luyi and Zhao Wenyuan.
  • Liu Dahai told Zheng Xiaobo that the other model had outperformed Kun Yuan on live tests and scored 85.2 on intent understanding compared with Kun Yuan’s 78.
  • Liu Dahai explained that the tests came from varied sources and that the other team’s benchmark covered 1,200 questions across sixteen scenarios, making fraud and overfitting unlikely.
  • Liu Dahai ruled out an architectural breakthrough and concluded the model’s performance likely came from superior labeled data.
  • Zheng Xiaobo listened to Liu Dahai’s analysis and said, “ScaleAI.”
Read chapter

Ch. 145Poach His Data Supplier

Zheng Xiaobo ordered Dingsheng to investigate and poach Source Code Technology’s data supplier so Kun Yuan could obtain Glutinous Rice Ball’s high-quality training data. Meanwhile, Han Luyi arranged to meet his old friends after Zhang Haoran asked for help with his work troubles.

  • Liu Dahai explained that Glutinous Rice Ball’s strong performance came from exceptionally precise annotation, and Zheng Xiaobo ordered Cheng Yuan to investigate Source Code Technology’s data supplier and poach them.
  • Zheng Xiaobo told Cheng Yuan to discuss a computing-power partnership with Han Luyi, but require data sharing and delay responding for two days.
  • Liu Dahai said fifty thousand samples annotated at that precision could make Kun Yuan the world’s leading model.
  • Han Luyi finished annotating more than three hundred data entries, then caught up on messages from Ma Xiaofei and Zhang Haoran in the Dorm 404 group.
  • Han Luyi offered to treat his friends that weekend, and Zhang Haoran enthusiastically agreed to meet.
Read chapter

Ch. 146So You're the One Doing Localization?

Jiang Yixin took charge of coordinating Source Code Technology’s Open Things localization project with NeusAI and began studying the Mercury model and American users’ needs. After Qian Xiaole described Mercury’s automated workflow, Jiang Yixin suddenly realized something important.

  • Jiang Yixin became the project’s de facto lead and arrived early to plan the team’s work on Mercury, product localization, and coordination with NeusAI.
  • Jiang Yixin reflected on Airbnb’s failure in China and wrote that the team should build a tool Americans would use, not merely an English version of Open Things.
  • Jiang Yixin struggled to follow Sarah’s rapid speech in video meetings but relied on AI transcription and translation to coordinate with NeusAI.
  • Jiang Yixin invited Qian Xiaole and Li Ting to lunch, where she asked Qian Xiaole about Mercury’s capabilities.
  • Qian Xiaole described Mercury’s automated workflow and complained that it repeatedly asked for approval, prompting Jiang Yixin to jerk her head up with a sudden realization.
Read chapter

Ch. 147Since it's so smart, let's call it an Agent

Jiang Yixin and Qian Xiaole built a rough internal tool that used AI to interpret tasks, edit code, run tests, and fix issues without human approval at every step. After naming it Agent, Jiang Yixin sent Su Niannian a demo, and Su Niannian hurried over to see it.

  • Jiang Yixin realized that repeated human confirmations were unnecessary when users were not reviewing the AI's work, and proposed building a tool that could complete tasks directly.
  • Jiang Yixin and Qian Xiaole designed a continuous workflow connecting the Mercury Model, Heavenly Craft, and BugKiller, then divided up the implementation.
  • Qian Xiaole directed AI to write much of the code while she handled the calling logic, and they created a functional demo in two hours.
  • Jiang Yixin tested the demo by asking it to add a dark mode toggle, and the tool updated the project, passed tests, and found no bugs in about thirty seconds.
  • Jiang Yixin and Qian Xiaole named the tool Agent, and Jiang Yixin sent Su Niannian a video demonstrating it.
  • Su Niannian sent Jiang Yixin a voice message asking her to stay at the office so she could come see Agent.
Read chapter

Ch. 148Wait Until I Give It a Better Brain (Bonus chapter for alliance leader [Spellcaster])

Jiang Yixin demonstrated an AI coding tool she had built with Qian Xiaole, and Su Niannian and Han Luyi tested it on a complex refactor that it could not complete. Han Luyi approved its internal rollout and, recognizing that the model’s capabilities were the main limitation, prepared to pursue a collaboration with Cheng Yuan to give it a better brain.

  • Su Niannian asked Jiang Yixin to demonstrate her AI coding tool live, then tested it on a core Open Things refactor that failed after ten retries.
  • Su Niannian brought Jiang Yixin and her laptop to Han Luyi, who examined the logs with Sight and concluded the model lacked the capability and training to complete the complex task.
  • Han Luyi praised Jiang Yixin’s work and told her to refine the tool with a small team, use it internally, and collect feedback, especially reports of failures.
  • Su Niannian told Jiang Yixin to prioritize the new tool while continuing to handle the collaboration with Neus.
  • Han Luyi reviewed Cheng Yuan’s message about collaborating on his demonstration prototype and sent his previously drafted terms.
Read chapter

Ch. 149What Did You Find?

Cheng Yuan's investigators failed to find any supplier or labeling team connected to Source Code Technology, leaving him to wonder whether Han Luyi had labeled the training data himself. A message from Zheng Xiaobo's secretary pressed him for a conclusion, and Cheng Yuan ended the chapter holding a letter of intent for computing power cooperation with Source Code Technology.

  • Wu Youtian and President Zhou reported that their investigations found no evidence Source Code Technology had hired data labelers or purchased labeling services.
  • President Zhou suggested Source Code Technology might have labeled its own user data, and Cheng Yuan challenged how the company had obtained information about intent-understanding data.
  • Supervisor Chen said his classmate had seen Han Luyi working in his office from morning until night, but had heard of no data-labeling team or other model projects.
  • Cheng Yuan angrily dismissed the investigators and wondered whether Han Luyi had labeled all the data himself.
  • Zheng Xiaobo's secretary asked whether Cheng Yuan could bring a conclusion to Wednesday's review, and Cheng Yuan picked up a letter of intent for computing power cooperation with Source Code Technology.
Read chapter

Ch. 150One More Skewer

Han Luyi met Zhang Biao, Ma Xiaofei, and Zhang Haoran for dinner, where Ma Xiaofei's gossip about a lawyer leaving Boheng made Han suspect the story involved Gu Siyue. Afterward, Han invited Zhang Haoran to try his new conversational model, and Zhang studied the login details Han sent him.

  • Han Luyi joined Zhang Biao at One More Skewer and used Sight to see that Zhang Biao felt regret.
  • Ma Xiaofei arrived recording with a DJI Osmo Pocket 4P, and the group ordered skewers and drinks.
  • Ma Xiaofei described a female Boheng lawyer who left after clashing with the firm's new director, making Han wonder if she was Gu Siyue.
  • Zhang Haoran recalled meeting a female Boheng lawyer through Han, but Han said she had never told him about the dispute.
  • Ma Xiaofei secured a twenty-percent discount by telling the owner that Han Luyi's company had built the shop's mini-program.
  • Han arranged to help Zhang Haoran open an account on Monday and sent him access details for a closed-beta conversational model.
Read chapter

Ch. 151Expand to View Thought Process

Zhang Haoran used Glutinous Rice Ball to assess the political struggle over replacing the bank’s core systems with domestic alternatives. The AI recommended that he submit a technical report supporting phased migration, and he finished it before wondering why Vice President Liu opposed the project.

  • Zhang Haoran avoided Xu Jing and brooded over the upcoming meeting about domestic replacement of the bank’s core systems.
  • Zhang Haoran recalled how Vice President Liu had opposed migration, Assistant Jiang had supported it, and President Guo had remained silent at their earlier meeting.
  • After Ma Xiaofei sent him a video, Zhang Haoran logged into Glutinous Rice Ball and asked what he should do.
  • Glutinous Rice Ball asked about Liu, Jiang, and Guo, then advised Zhang Haoran to submit a technical assessment recommending phased migration without openly declaring his position at the meeting.
  • Zhang Haoran expanded the AI’s thought process, concluded that headquarters and President Guo were likely behind Assistant Jiang, and completed a feasibility report supporting migration.
  • Zhang Haoran removed “First Draft” from the report’s title and wondered why Liu was opposing a project that could bring results and credit.
Read chapter
Arc 6 of 21
AboutStatsAPIChangelogStatusContactPrivacyTermsDMCA
© 2026 RandomTranslator·v3.26·RSS