Upon hearing Qian Xiaole's words, Jiang Yixin, who had been feeling somewhat tense, suddenly found herself at a loss for words, feeling both amused and exasperated.
"Don't change your nickname randomly," Jiang Yixin said. "What if Riding the Wind becomes popular later? Wouldn't that be considered riding on his coattails?"
Qian Xiaole stated confidently, "This is my behavior as a spiritual shareholder, a co-creator of the product."
Li Ting chimed in from the side, "You two, stop talking about the game for now, the landing page isn't finished yet."
Jiang Yixin then turned her attention back to the screen.
She replaced "programming Agent" with "Riding the Wind" on all the pages.
"Source Code Technology - Riding the Wind Closed Beta Invitation."
"Assign repetitive, tedious, and low-value development tasks to Riding the Wind."
It automatically breaks down tasks, generates code, runs tests, and continues to fix errors based on the results.
The current version is recommended for low to medium complexity development tasks, but not for high-complexity cross-module refactoring.
Jiang Yixin looked at the last sentence, thought for a moment, and then kept it. Just as Su Niannian said, managing user expectations is also part of the product.
After completing all the preparations and confirming everything several times, Jiang Yixin went home after work.
Friday morning, 9:00 AM.
Jiang Yixin sat in front of the computer, her eyes fixed on the screen, her hands clenched into fists.
The beta landing page went live on the official website on time, and in-app messages with pre-set timed tasks were sent to the designated users on the list. There were two channels: fifty users were distributed through the official website, and fifty were sent via targeted invitations.
Not only Jiang Yixin, but even Qian Xiaole arrived at the office early, which was unprecedented for them.
Invitation codes claimed: 0.
Activation count: 0.
Waiting queue: 0.
I watched for two minutes, and the numbers didn't change.
Jiang Yixin knew the webpage was updated in a streaming manner, but she still couldn't help refreshing it.
The backend page still shows zero.
Jiang Yixin and Qian Xiaole exchanged a glance. Qian Xiaole said, "Could it be that there's a delay in the internal messaging system? Send me one to try."
Jiang Yixin then switched to the system management page of the site's internal messaging system, manually entered Qian Xiaole's ID, and sent an invitation.
Qian Xiaole picked up her phone, glanced at it, and said, "Received it. Maybe everyone's going to work late on Friday."
Suddenly, she pointed at the screen and said, "Look, it's here!"
Jiang Yixin turned around and saw that the data in the backend had changed.
Number received: 1.
Then, the second one.
The third one.
By 9:23, all fifty invitation codes on the official website had been distributed.
The number of activations in the backend reached sixty-seven, which is the combined number from official website invitations and targeted invitations. At the same time, the number in the waiting queue also began to rise, representing users who wanted to participate after the official website invitation codes ran out.
Based on the user agreement for the internal testing, detailed call information for Agent calls during the internal testing period can be viewed in the background.
Jiang Yixin glanced at the access list and breathed a sigh of relief.
Early adopters were also cautious; no one immediately released production project code to this new tool.
Most of the projects we connect to are public repositories or open-source projects on GitHub, which anyone can access.
About five minutes later, most of the first batch of users had completed their initial experience with the Agent. Jiang Yixin could see from the backend that the first-time pass rate was around 80%. That is to say, after the first round of use, approximately 80%...
Users chose to keep the code changes instead of reverting to the previous version with a single click.
Qian Xiaole leaned closer. "These data look pretty good."
"Eighty percent is acceptable," Jiang Yixin without taking her eyes off the screen. "The internal data is over ninety percent."
"Isn't that too low?"
"The pass rate only increased after internal staff used it for a while; there was a learning process," Jiang Yixin said. "The beta testers are using it for the first time, and 80% of them have already exceeded expectations."
Qian Xiaole nodded. "This shows that our recent iterations are still effective; we can call it a good start."
"Mm," Jiang Yixin responded.
"So what do we do next?" Qian Xiaole turned around and asked. It was her first time participating in a product launch like this, and she was unsure of what to do.
"Wait for the data, wait for the feedback," Jiang Yixin said. She had written down the next steps after the release in a document she had prepared in advance.
Logically, at this point, with the internal testing confirmed to be going smoothly, Jiang Yixin could go and attend to other tasks.
After all, this is just a closed beta test, so the impact, whether good or bad, is very limited. There's no need to keep a close eye on it like when a new version of Open Things is released.
However, Jiang Yixin couldn't help but keep staring at the data. Every time the number of calls and waiting times increased, she felt a joy she had never experienced before.
Those are all real users, using or queuing to reserve the products she makes.
This feeling is addictive.
Jiang Yixin just stared at it, forgetting even to drink her water.
When Li Ting came to call her for lunch, she looked up and realized it was already noon. She tidied herself up and stood up.
Li Ting went to call Qian Xiaole again, Qian Xiaole waved her hand and said, "You guys go ahead, I brought sandwiches today, I'll eat them in the office."
Walking down the street, neither of them spoke; Jiang Yixin was still replaying the data from the morning in her mind.
All one hundred invitation codes sent out have been activated, with a total of two hundred and seventy-nine calls. The user who used the most codes has called them eleven times.
Keep in mind that only three hours have passed, and this is during working hours. Each time Riding the Wind invokes the entire process, the length varies, but on average it takes about five minutes. Invoking it eleven times can be considered an addictive use.
In terms of user retention rate alone, this figure is already quite remarkable.
When they arrived at the fast food restaurant, Jiang Yixin's smile was still on her face. The cashier at the fast food restaurant couldn't help but smile at her when she saw her.
Sitting down at the dining table, Li Ting waved her hand in front of Jiang Yixin's eyes. "President Jiang, tone down your smirk. Is the beta testing going that smoothly?"
Jiang Yixin rubbed her face. "Was my behavior that obvious?"
"It's super obvious."
After dinner, Jiang Yixin and Li Ting went to buy milk tea together, and also bought a cup for Qian Xiaole, who couldn't come.
As soon as I got back to my workstation, I saw Qian Xiaole looking down at her phone, her expression off.
"President Jiang, President Jiang, come quick!" Qian Xiaole handed over her phone when she saw Jiang Yixin return.
Jiang Yixin took it; it was a popular post on a tech forum.
I tried out Source Code Technology's "Riding the Wind," and here's why I don't recommend it to anyone.
"I'm a long-time user of BugKiller and Open Things. This time, I was invited to test Source Code Technology's new product, Riding the Wind, and I immediately conducted an in-depth test. My conclusion is this: Riding the Wind is not a good tool; it does more harm than good to the industry's development."
In terms of user experience, it's really fast at completing simple tasks. Tasks like patching test cases or modifying interface parameters, which might take me a whole morning to do myself, can be completed in just a few minutes with Riding the Wind.
But I have to tell you something that makes me shudder.
I specifically tested a slightly more complex task: refactoring a payment module. The Riding the Wind ran three times, taking about twenty minutes, and generated a large amount of code. All changes included unit tests, and the final result was all green (positive feedback), with a checkmark next to each test.
To be honest, I almost used it directly.
But since it was a test, I forced myself to read it once.
I was quite surprised when I first saw it; the code structure was really beautiful, the variable names and comments were even better than what a real person would write.
However, there is a piece of logic within it that it made up itself.
The payment channel I'm using has a very strange behavior: if I retry after a timeout, it doesn't return a "duplicate request" error, but instead deducts the money again. In short, they don't do deduplication on the Server side. I've encountered this situation before; a customer was charged twice and almost drove me crazy. So I had to write my own deduplication logic.
However, it was removed during the reconstruction of Riding the Wind without any prompts or questions.
This is the biggest problem with AI tools: they don't know what they don't know.
If you give this task to any new programmer, they might be slow or unable to complete it, but they would never delete a piece of code running in the production environment without even asking any questions.
So I say, the biggest problem with Riding the Wind is not that it's not easy to use, but rather that it's too easy to use.
It's like Pandora's Box; once opened, someone will inevitably be tempted to use it to write code. However, the resulting code will contain mostly correct logic mixed with "illusions" that appear out of nowhere.
Will the entire edifice of the internet be built on such an unstable foundation?
As a veteran in the industry, I advise every programmer not to use Riding the Wind.
As a long-time user of Source Code Technology, I advise you not to release this product.
Above."
The poster is CodeMonkey Old Zhang, the same person who said it had "no bottom line" when it All Things Alive. Old Zhang is a tech influencer on the internet, and the post already had hundreds of replies in less than an hour, with some people even tagging the official account of Source Code Technology in the comments.
Jiang Yixin had just finished reading when she looked up and saw Little Yang rushing over to find Li Ting in a panic.
"Team leader, come quick! Our official account's new post from this morning is getting flooded with comments!"
Li Ting walked quickly back to her workstation, sat down, turned on her computer, refreshed the page, and her face darkened.
Before she could even see Old Zhang's post, what she saw was that the official account of Source Code Technology was already in an uproar.
The comment section below the announcement of the Riding the Wind posted this morning has completely spiraled out of control.
The comments under the post were normal in the morning, with messages like "Claimed it, looking forward to trying it out" and "Report from a long-time BugKiller user."
Now all those comments have been pushed down.
The top comment is a link to CodeMonkey Old Zhang, followed by the sentence, "Does Source Code Technology want to destroy the livelihoods of all programmers!?"