APTnovel Chapter 493 Short-term Results
Chapter 493 Short-term Results
Two weeks later, TUTU held an "Allies Data Sharing Meeting".
JD.com, Ele.me, Kuaishou, and TUTU themselves will share their achievements after using artificial intelligence.
The conference room was packed.
On either side of the long table were the core leaders of the three allies. Zheng Kai from JD.com brought two vice presidents, Liu Chengdong from Ele.me brought a technical director and an operations director, and Zhang Yifan from Kuaishou and Douyin did not come. Instead, Li Mingyuan, the head of the recommendation algorithm group, and a business director came.
Zhou Mingzhe, Chen Mo, and Zhao Yiming were all present at TUTU's side, while Lao Feng sat by the window flipping through materials.
The projector was on, and several printed data reports were spread out on the table, the papers fluttering gently in the air conditioning breeze.
Lu Ran sat at one end of the long table, glanced around at everyone present, and spoke in a relaxed tone, as if inviting friends over for a casual meal: "Everyone, we've been working on this collaboration for over half a month. Let's not talk about formalities today. Everyone, lay out the facts of this period, good or bad, let's speak frankly."
Zheng Kai was the first to respond.
He turned the laptop around, and the screen displayed a line graph. The curve had climbed up a slope that wasn't steep but was quite noticeable from its starting point a month ago.
He pointed to the line, speaking at a measured pace: "Let me first tell you about our data. After the AI supply chain module went live, we tested it for three weeks in the fresh produce and daily necessities categories, and the results were much better than we had internally anticipated. The stockout rate for fresh produce dropped by 15%, and the unsold rate dropped by nearly 20%. The results for daily necessities were slightly worse, but there was still a significant decrease. Overall, these two categories saved us nearly two million yuan per month in warehousing and spoilage costs."
He paused, then turned to the next page, this time a bar chart: "And it's not just about cost improvements. User experience data is also improving. Out-of-stock situations are less frequent, and the probability of users being notified 'this item is temporarily out of stock' after placing an order has significantly decreased. Customer service has received about 15% fewer complaint calls than last month, and user satisfaction scores have also slightly rebounded."
A JD.com vice president nearby added, "Moreover, the internal resistance to implementation was less than expected. After seeing the first batch of data, the heads of various business lines proactively asked when it could be rolled out to their areas; we didn't need to push it any further. Starting next week, we plan to expand the AI module to all self-operated product categories."
Liu Chengdong took it after Zheng Kai finished speaking.
He projected the data from his phone onto the screen, his tone carrying the lightness that comes with receiving good data: "We've also run two rounds of testing. The first round was a pilot program in a medium-sized city, where the average delivery time per rider decreased by eight minutes, and the empty-run rate dropped by about 20%. The second round expanded to two more cities, and the results were basically the same, with no deviations. Now, the average delivery cost per order is almost one yuan lower than before the integration."
He put down his phone, crossed his hands on the table, and his expression became more serious: "One yuan may not sound like much, but multiplied by our daily order volume, and then multiplied by a month, it becomes a considerable sum. Moreover, we didn't convert all of this cost saving into profit—we used a portion of it for user subsidies, which further boosted order volume. Simply put, increased efficiency leads to lower costs, lower costs create a price advantage, and a price advantage attracts more users. It's a positive cycle."
Li Mingyuan sat on the other side of the fast-paced video platform and waited for the two people in front of him to finish speaking before speaking.
His style was more technical than Zheng Kai's and Liu Chengdong's, and he habitually drew circles on the table with his finger while speaking: "The data from Kuaishou and Douyin is a bit more complex. Our recommendation system itself has already reached a very high level, and there's not much room for further optimization on this system. But after the AI engine was integrated and ran for two weeks, the average user time increased by nearly two minutes. This number sounds small, but it's an increase from an already high base, which is much more difficult than going from zero to thirty minutes."
He paused, then added, "Moreover, this growth hasn't shown any obvious signs of slowing down. According to our internal model, if we optimize it for another one or two versions, this growth could increase even further."
After the three companies finished speaking, there was a subtle change in the atmosphere of the conference room.
No one deliberately applauded or said anything nice, but the "not bad" expression on the faces of the people sitting at the table slowly turned into an "actually good" expression.
Some people scribbled notes in their notebooks, some took a sip of water and put it down, and some leaned back in their chairs and stared silently at the data on the projector for a while.
Lu Ran waited a few seconds before speaking, "All three companies have presented quantifiable results. From this perspective, the commercial value of AI technology has been validated by actual data. Now I'd like to discuss a more specific question—besides the data, have there been any other noteworthy developments recently?"
Zheng Kai spoke first: "We've received inquiries from some peer companies. Two e-commerce companies contacted us through third-party channels, asking if they could also use TUTU's AI solution. They didn't contact you directly, probably because they felt it would be easier to ask through us as an intermediary. I can't name the two companies right now, but they're both quite large; one is a vertical e-commerce company, and the other is a cross-border business."
Liu Chengdong nodded in agreement: "Ele.me is experiencing a similar situation. Two local service platforms have proactively inquired about cooperation—one specializing in community group buying and the other in housekeeping services. They've both seen the improvements we've made in delivery efficiency and want to know if the same technology can be applied to their own dispatch systems."
Li Mingyuan concluded by saying, "We've received relatively few inquiries from Kuaishou and Douyin because we haven't publicly disclosed many technical details about our recommendation system. However, an online education company contacted us, saying that their content recommendation system has been performing poorly, with users leaving before finishing the courses. They asked if we could introduce TUTU's AI solution to them. I think this direction is quite interesting. Although educational content recommendation and short video recommendation are not exactly the same, their underlying logic is similar."
As Lu Ran listened, he silently memorized the names of these potential partners.
However, he did not rush to express his opinion, but simply nodded, saying that they could maintain contact first and not rush to move forward.
Next up is TUTU's report.
Chen Mo stood up and projected the data from TUTU's internal AI applications onto the screen.
His tone was more even than the previous speakers, as if stating a settled fact: "TUTU benefits from AI technology more broadly than our partners. Firstly, there's the improvement in development efficiency. Zhao Yiming did a survey last month; creating an operational page for a medium-sized event, which used to take three to five days from design to development to testing and launch, has now been reduced to less than a day and a half. This isn't achieved by increasing manpower, but by automating all the repetitive tasks through AI-assisted code generation."
He turned a page and continued, "Secondly, there's the efficiency on the content side. Previously, for content like match previews and reports, editors had to write everything from scratch. Now, AI helps generate drafts, and editors only need to revise and polish them. Production time for similar types of content has been reduced by more than half. The content team hasn't increased in size, but output has doubled."
"And then there's customer service. User inquiries have been steadily increasing, but the customer service team hasn't undergone a large-scale expansion. The AI-powered automated response system handles about 60-70% of routine inquiries, while the remaining 30% requiring manual processing are transferred to human agents. The workload for the customer service team hasn't increased due to user growth; in fact, it's become somewhat lighter than before."
After Chen Mo finished speaking, Zhao Yiming added, "The most obvious issue is manpower. Previously, the company had a recruitment shortfall of about several dozen positions, which we couldn't fill because suitable candidates were hard to find on the market. But after the AI tools were implemented, one person can do the work of one and a half to two people, significantly reducing the recruitment pressure. Some positions still haven't been filled, but the work progress hasn't been slowed down at all. In a sense, AI has helped us alleviate our manpower problems."
Upon hearing this, Lu Ran silently estimated the total increase in overall efficiency.
Game development, content creation, customer service, data analysis, and operational strategy development—every aspect is being reshaped by AI tools to redefine the efficiency ceiling.
Adding all of this together, TUTU can now do about one and a half times the work it used to with the same number of people.
The meeting ended at around 4:30 p.m.
Before leaving, Zheng Kai shook hands with Lu Ran at the door and said, "This meeting was worthwhile." Liu Chengdong chimed in, "When you get back, compile a list of today's data for me, and I'll take it back to show our finance department colleagues." Li Mingyuan didn't say much, just nodded and said, "Notify us in advance when there are subsequent version updates."
After the three people left the meeting room with their respective groups, the corridor quickly became quiet.
Zhou Mingzhe was the last to leave. He paused at the door, turned around, and said, "Today's meeting was more effective than ten strategy meetings. The data speaks for itself; they can see the value of the cooperation without us having to repeatedly prove it."
Lu Ran nodded but didn't reply.
He returned to his office, sat down, and mentally reviewed the data he had heard at the meeting earlier that day.
JD.com's supply chain efficiency, Ele.me's delivery costs, and Kuaishou and Douyin's user time—three completely different industry application scenarios—have all yielded quantifiable positive results.
This means that the versatility and adaptability of AI technology are even greater than he had anticipated.
As for potential partners who actively seek them out—vertical e-commerce, cross-border business, community group buying, housekeeping services, and online education—each direction corresponds to an independent application scenario.
If these companies can all access AI capabilities through cooperation, then TUTU's network of allies will no longer be limited to the three scenarios of "shopping, food delivery, and short videos," but will extend to more niche areas.
He leaned back in his chair and thought for a moment, then opened his notebook and wrote a line at the end of that day's page: "The commercialization validation phase of the AI engine has been completed. Entering the large-scale expansion phase."
After finishing writing, he closed the notebook, stood up, and stood by the window for a while.
Pedestrians dressed in spring clothes strolled along the street downstairs, and the paulownia trees along the roadside were covered in pale purple flowers, the petals of which were blown down by the wind and carpeted the ground.
He looked at the flowers for a while and felt that the morning had been much more fulfilling than he had expected.
After the three companies returned with impressive data, these achievements gradually took hold in their respective industries.
Over the next two or three weeks, the list of partners contacting TUTU through JD.com, Ele.me, Kuaishou, and Douyin continued to grow, covering industries ranging from e-commerce delivery to content distribution and online education, and then to several more vertical fields.
Qunar Travel was among the first to formally connect with TUTU. After contacting Chen Mo through Kuaishou and Douyin, they quickly assembled a technical team and held two rounds of online meetings with TUTU.
The people at Qunar are very direct. They say their biggest pain point is the short user dwell time. Users search for flights and hotels and then leave, using the service and leaving without any residual value.
If AI recommendations can make users take a second look at nearby travel options and browse more destination guides after booking their trips, then this incremental growth is real and tangible.
After talking with the other party, Chen Mo gave Lu Ran feedback—Qunar was very willing to cooperate, and their business scenarios and Kuaishou's recommendation scenarios had a lot of reusable logic. If the framework agreement could be finalized, the speed of technical integration would be faster than the previous companies.
The ride-sharing service also sent a clearer signal through an intermediary.
The intelligent dispatch team within the ride-sharing service is very interested in Ele.me's delivery data because they are also doing similar optimizations themselves—how to ensure that each vehicle picks up the nearest passenger in the shortest time, how to predict ride-hailing demand in different areas, and how to reduce empty driving time.
These issues are highly similar to the fundamental logic of food delivery, only the application scenario has changed from delivering food to delivering people.
The ride-sharing company's attitude has become much more proactive than before. Although negotiations have not yet officially begun, the technical teams of both sides have established a preliminary communication channel.
The ride-sharing company said they need to conduct an internal assessment first to confirm the compatibility of the AI dispatch system with the ride-sharing company's existing technology before they can move to the next step.
This statement wasn't quick, but it represents a significant step forward compared to the previous wait-and-see approach of "let's find out first."
Another cross-border e-commerce company also expressed its intention to cooperate through JD.com.
This company is not particularly large, but it has a presence in several markets in Southeast Asia and Latin America. The core problem it faces is the same as JD.com's – the management cost of cross-border supply chains is too high and the inventory turnover efficiency is too low.
After seeing JD.com's data, they felt that this solution could also be used in their own business, so they proactively asked TUTU if they were interested in adapting it for cross-border scenarios.
Lu Ran had Zhao Yiming conduct a technical assessment first, and the results showed that the supply chain optimization logic of cross-border e-commerce is more complex than that of domestic e-commerce—involving multi-country warehousing, cross-border logistics, and the impact of different tariff policies, with variables an order of magnitude greater than those of a single market.
However, he did not immediately refuse, but instead told Zhao Yiming to keep this direction and consider expanding it after the main battlefield was stabilized.
These potential partners have one thing in common—they didn't approach TUTU proactively; they came to us after seeing the effectiveness of AI technology through data from existing partners.
This means that the commercial value of the AI engine no longer needs to be marketed by TUTU itself; the data speaks for itself.
Lu Ran asked Chen Mo to compile these contacts into a list, categorizing them according to three dimensions: industry, strength of cooperation intention, and difficulty of technology adaptation.
The list contains about a dozen companies, of which seven have already been in initial contact, while the rest are still in the observation or internal evaluation stage.
After reviewing the list, he put it in his drawer, knowing that the next priority was not to continue expanding the list, but to move the companies he had already connected with to the stage of substantive cooperation.
No matter how long the list is, without a concrete agreement, it's just names on paper.
He picked up his phone and sent Chen Mo a message, simply saying, "Arrange for a formal meeting next week."
Chen Mo replied quickly: "It's already arranged. Tuesday afternoon at 2 PM."
...