Vibe Coding

Su Niannian stared at the user growth curve on the screen, unconsciously twirling a pen between her fingers.

The curve was still rising, but not fast enough.

BugKiller had been live for four months and had over two hundred paying teams. That was not bad for a startup, but she felt they should have done better given the strength of their product. That was her responsibility as COO.

The real pressure came from CodeSafe. It was built into Dingsheng Cloud and free, making BugKiller's pricing extremely awkward. CodeSafe had not only taken away Dingsheng Cloud's users, but also anchored the price expectations of users on other cloud platforms.

BugKiller's image had become: useful, but a little expensive. For most teams that did not absolutely need it, they would rather not buy it.

This was no longer a technical problem.

It was a problem of user psychology.

Jiang Yixin walked over carrying her MacBook, her face practically screaming, "I'm excited."

"President Su, I have an idea about that batch of anomalous users from this afternoon. Shall we discuss it in the meeting room?"

"Sure." Su Niannian got up and picked up her laptop as well.

Once they reached the meeting room, Jiang Yixin set her computer on the table. A CSV file was open on the screen: user IDs, submission times, code snippets, error types, all kinds of raw data piled together in dense rows.

"I ran through it the way you said." Jiang Yixin pointed at the screen. "There really is a problem. The code from this batch of users—the variable names, indentation styles, even their error patterns—are highly consistent. They are completely unlike the individual habits of normal programmers. I think their code was generated by AI."

Su Niannian glanced over the spreadsheet without looking at the details, then asked first, "How much data?"

"They account for 17% of total active users, but contribute 32% of submissions."

"Good." Su Niannian opened Lark. "Next time you give me data, do this first—"

She dragged the CSV into a Lark multidimensional table, selected several columns for a pivot analysis, and the system automatically generated charts. Thirty seconds later, three panels appeared on the screen: a submission frequency trend chart, a pie chart of code feature distribution, and a heat map of user activity times.

Jiang Yixin froze for a moment.

"Raw data is just raw material," Su Niannian said. "But the human brain cannot process thousands of rows in a spreadsheet. The first step in Business Intelligence, or BI, is turning data into patterns people can understand at a glance. Time distribution, behavioral clustering, anomaly tagging—visualization is not for looks. It is so decision-makers can immediately grasp the important points."

She pointed at the heat map. "Look, the concentration in this time period is much more intuitive than scrolling through a thousand submission records in a spreadsheet."

Jiang Yixin stared at the screen, her expression shifting from confusion to realization. On her first day at work, she had learned her true first lesson.

"Now that we have a visualization dashboard, what comes next?"

Jiang Yixin blinked. "What?"

"When you look at user data, do not only look at what they do. Look at who they are—that is called a user profile."

Su Niannian pulled up the section with users' registration information and tapped it with her finger.

"Look at the suffixes on their registration emails."

Jiang Yixin leaned closer, her gaze sweeping quickly across them.

"...Mostly company emails, but not internet companies. There are consumer goods companies, GG companies, and... real estate?"

"Now look at the distribution of submission times."

Jiang Yixin switched to the time heat map.

"Concentrated from two to five in the afternoon on workdays? That's not—"

"Not a programmer's work rhythm," Su Niannian picked up. "Normal developers are usually at their best in the morning, with a submission peak from ten to twelve. There is another wave around three or four in the afternoon. This group peaks around one or two at noon, and it lasts only a short time. It looks like they are doing it during their lunch break."

Jiang Yixin opened the code samples users had shared.

"President Su, look at this comment: 'After the user clicks the button, display a confirmation dialog. After confirmation, call the backend API to submit the order.'"

Su Niannian took one look.

"That is not a code comment. That is a product requirements document."

"Exactly! And this one: 'Display a loading animation when the page loads, then replace it with a list when the data returns.' People writing code would not write comments like that."

Jiang Yixin spoke faster and faster.

"Now look at the error types: undefined variables, mismatched brackets, misspelled function names. They are all basic syntax problems. Even a bad programmer would not make these mistakes so densely, unless..."

She stopped and looked up at Su Niannian.

"They are not programmers at all."

Su Niannian said nothing, but silently praised her. By the time she guided her to the third step, Jiang Yixin had connected the dots herself. She was quick.

"Then who are they?" Jiang Yixin asked herself before digging into the job title field in the registration data.

Five minutes later, the two women fell silent before the pie chart on the screen.

Product managers, 28%; operations, 22%; UI/UX designers, 18%; marketing, 12%; other non-technical roles, 20%.

Not a single programmer.

"They use AI to write code," Jiang Yixin said slowly, "then throw the bugs AI cannot handle to BugKiller to fix."

Su Niannian stood up and walked to the whiteboard.

"They are not just fixing bugs. Think about it—why do these people use AI to write code?"

"Because... they have ideas, but cannot program?"

"Right. Product managers want to build prototypes to validate requirements, operations staff want to make campaign pages, and designers want to turn mockups into interactive demos. In the past, all of this had to be scheduled and handed to developers. Now AI lets them do it themselves."

Jiang Yixin followed up, "But AI-generated code is not always right. When it runs into bugs, they cannot understand them."

"So they use BugKiller."

Su Niannian drew a line across the whiteboard. On the left, she wrote "AI Generation"; on the right, "Bug Fixing"; in the middle, she drew a stick figure.

"These people are not programmers. They are people who use AI to create products. Programming is not their job—it is a means of bringing their ideas to life. AI has lowered the barrier, but not to zero. That gap in the middle is our opportunity."

She wrote two English words in the middle of the whiteboard.

Vibe Coding.

"Vibe Coding?" Jiang Yixin read aloud.

"That's right. You do not need to understand every line of code. You create based on instinct and ideas. AI handles the writing, and when something goes wrong—"

"We catch them when they fall!" Jiang Yixin answered before she could finish. "Then what happens when they write something wrong? Wouldn't they have to come to us to fix it? This is technical insurance for ordinary people."

Su Niannian turned to look at her.

That was a good analogy—insurance gave people a sense of security. Users did not need to understand what an undefined variable was. They only needed to know that if something went wrong, someone had their back.

"Exactly. That is our value."

Her phone rang.

Caller ID: Mom.

Su Niannian glanced at the whiteboard, apologized, and stepped outside to answer it.

"Niannian, are you free this weekend? Auntie Li introduced a young man. He works at a bank and has quite good prospects. Go meet him?"

"No. I'm busy."

"You're busy, busy, busy every day. Busy until when? You're already twenty-seven. If you keep putting it off, all the good ones will be taken, you know?"

Su Niannian switched the phone to her other ear.

"Your dad said that if it really does not work out, you can bring back that out-of-town boy from back then for us to meet."

Su Niannian's fingers tightened.

"You were the ones who would not allow it back then."

There was a pause on the other end.

"Back then was back then. Why don't you bring a boyfriend home?"

"Out-of-town, no house, no local residency permit—we are not picky about any of that anymore." Her mother spoke faster. "As long as you are willing to find someone. As long as you are not alone."

"Mom, I'm in a meeting."

After hanging up, Su Niannian took a deep breath.

Jiang Yixin kept her head down and flipped through the data, tactfully pretending not to notice.

"Where were we?"

"Technical insurance."

"Mm." Su Niannian picked up a marker and drew three arrows beneath Vibe Coding. "This market has three characteristics. First, the user base is huge. Anyone with an idea is a potential user. Second, they have a strong willingness to pay. What they are buying is not technical service, but time and certainty. Third, it is growing fast. The more widespread AI tools become, the larger this group will grow."

She looked at Jiang Yixin, who was taking notes at high speed.

"Come on. Let's go find President Han."