Recently, chatting with a few friends, plus spending ten-plus hours daily immersed in coding agents myself, one feeling has become increasingly clear: this wave of AI isn't one revolution—it's two.
The Programming Track Has Been Discussed to Death
The first track is programming and productivity. Coding agents, AI office assistants—the entire workflow of white-collar workers in cognitive processing and software creation is being reshaped. There's been plenty of discussion on this track, and I've written about it before, so I won't expand on it here.
Some numbers to give you a sense of scale: GitHub's statistics say AI has written 26.9% of production code, developers using AI save an average of 3.6 hours per week, and merged PRs are up 60%. But there are dissenting voices—METR's randomized controlled trial found that core contributors to 16 large open-source projects actually slowed down by 19% after using AI tools, though they felt 20% faster. Faced with complex, large codebases, AI hasn't yet reached the point where using it mindlessly is always better.
My own sense is that when you know what you want, coding agents can indeed boost efficiency by an order of magnitude. The key phrase is "knowing what you want."
The Other Track, Probably Underestimated
The second track is content creation. Text-to-image, text-to-video, image editing—what's happening in this direction is no less intense than the programming side.
ByteDance's Seedance 2.0 is a watershed moment.
Released in February, API opened at the end of March. Artificial Analysis benchmarked it at Elo 1269, surpassing Google Veo 3, Sora 2, and Runway Gen-4.5. Not just a little ahead—it's in a different league. It can generate up to 20 seconds of 1080p video in one go, with music, dialogue, and sound effects synchronized, no post-production dubbing needed. Camera movement, lighting, character actions—all precisely controllable.
A friend of mine working on AI short dramas said there are tons of teams in the industry sitting on tens of millions in RMB cash waiting to use this. The day the API opened, my social media feed exploded with entrepreneurs in related fields—short films made with it flooded the screen, and the results were completely different from before.
The cost changes are staggering. Previously, producing a 25-minute episode of Japanese animation cost roughly 1 to 3.2 million RMB (450k USD). Attack on Titan was about 1.1 million per episode, Jujutsu Kaisen about the same. Now a 3-minute AI short costs 400 to 1,200 RMB (170 USD). Per-minute costs have dropped to a few percent of traditional methods. In blind tests, 73% of viewers couldn't tell the difference.
I also have friends running small live commerce teams who tell me content production costs have dropped to one-tenth of before, and speed has increased tenfold. Previously a content person cost over 10,000 RMB monthly; now the same output costs an order of magnitude less. Completely supply-constrained.
Imagine someone spending a few thousand RMB and one week to make a 20-minute anime short. Previously impossible. Now the outline is clear—text-to-video has reached its "ChatGPT moment." Once the path is found, cost reduction is only a matter of time. That's AI's rhythm: someone blazes a trail, then cheaper solutions inevitably follow, because it's fundamentally software, endlessly iterable through data and training.
Two Paths Diverging
I have a friend whom I recommended coding agents to. He said he hasn't had the energy to look into them lately—not because he isn't smart, but because his time and passion are entirely in content creation, figuring out how to use the latest tools to build pipelines, how to express creativity at low cost. I thought about it—why should he research coding agents? That's not his direction.
The reverse is also true. I spend ten-plus hours daily in coding agents; asking me to research content creation pipelines, I couldn't reach that level in a short time.
In my previous post, I wrote about the divergence between Builders and Promoters. This time the feeling is more concrete: these two paths differ not just in roles, but in tools, skills, and passion required. There's overlap, but it's shrinking, while the divergence is growing.
This means it's not just product builders undergoing massive change. People passionate about expression—content creators—will next receive tools of equal magnitude. If you have strong desire to express something, for the cost of a few dozen RMB and half a day, you can make a one-minute short video. Smartphones and Douyin (TikTok) already lowered the barrier to shooting videos once; next, that barrier will drop another order of magnitude.
Agent to Agent: Meetings Can Die
Both paths share a common downstream impact: both are changing how people interact with each other.
A friend asked me: AI is so powerful now, so if I return to face-to-face interaction fields—sales, investor relations, supply chain—is that safer?
I don't think so necessarily.
I now think meetings are an extremely inefficient method. I record audio, AI transcribes it and organizes it into a document, with more information than a two-hour meeting. The other party uses an agent to understand this document, extracting the most crucial points in minutes. What I do in five minutes might take two hours of meeting to get the full picture.
The efficiency difference is 100x. This isn't rhetoric.
Infrastructure is rapidly taking shape. Google released the Agent-to-Agent protocol (A2A) last year, with over 50 enterprises including Salesforce, SAP, and PayPal pushing it forward. Anthropic's MCP lets agents access various tools and data sources. Early this year, the Linux Foundation established the Agentic AI Foundation, with OpenAI, Anthropic, Google, and Microsoft all joining; by February, over 100 enterprises followed. Gartner predicts that by year-end, 40% of enterprise applications will embed AI agents; last year that number was under 5%.
Your agent and the other party's agent analyzing, transmitting information, and coordinating tasks in the background will be far more professional than two people chatting face-to-face. Agents don't need small talk, don't need two weeks to build trust. They can synthesize judgments from all verifiable information, more reliable than listening to someone say two sentences and then "feeling like this person is okay."
What E-commerce Eliminated
This reminds me of e-commerce.
Before e-commerce, all transactions had to be face-to-face. To buy something, you had to meet the seller, establish a relationship, judge each other. The seller's social skills and charisma were key factors in closing deals.
E-commerce arrived. Products are similar, compare prices, look at reviews, order in ten minutes. You don't know the seller, haven't met them, they don't even know you bought. In B2B too—buyers complete 57% to 70% of procurement research before contacting sales, and 67% of the procurement journey happens online.
E-commerce didn't eliminate retail. US physical retail still accounts for 81.6%. But it restructured the relationship logic in transactions—face-to-face social skills went from "essential" to "nice to have."
Agent-to-agent will bring a similar but much stronger wave. When the efficiency difference is 100x, in many scenarios people will choose the more efficient method. It's not that people aren't important; it's that business matters now have better channels for handling.
Don't Talk Business on the Court
There's a side effect of this that I think is quite good: human relationships will become purer.
Before, you had to talk business when playing ball, discuss partnerships over meals. Face-to-face was the most efficient business communication method, so social and business were always mixed. You didn't go play ball just because you liked it, but because you could meet people you wanted to know on the court.
If agents solve the business part? Playing ball is just playing ball. Socializing is just socializing. No need to maintain unwanted relationships in situations you don't enjoy.
Intimacy, entertainment, interests—these needs will be separated from business scenarios. It's not that they aren't important; it's that they can finally just be themselves.
Go Find It
Two revolutions running simultaneously, each requiring massive time to master.
The pattern I've observed is simple: people with passion soak ten hours daily, and they master every tool. People without passion occasionally open and look, and the gap quickly becomes orders of magnitude. The AI leverage is there; whether you can pry it loose depends on whether you're willing to keep applying force.
Whether building products or creating content—go find that thing you can't stop doing.