At lunch today, I overheard most of a conversation from the table next to mine.
They were probably peers—there are a lot of tech companies in this area. The first half was all complaints about the changes AI had brought, saying software is becoming less valuable, that they now need to find two or three times as many clients to meet the same performance targets as before, and that they feel more exhausted.
Then suddenly the tone shifted. Someone suggested whether they could build a replacement for some HR-related service themselves. Another person immediately chimed in, saying they knew people with resources, and could this be turned into a business?
They had been anxious just moments before; now they were excited.
Anxiety and Excitement—Everyone Has Them
It wasn't just that table. In elevators, at restaurants—casual conversations these days are basically all about AI. It's the same with friends around me: anxious that their old rhythm has been disrupted, excited that there might be new opportunities.
These two emotions alternate. Everyone is the same. I think this is just the new normal.
Questioned After Nine Days Without a Release
The pace of change is indeed accelerating.
Today OpenClaw released their latest version. Before that, there had been no updates for nine days, and many in the community were asking: What's going on? Are they cooking up something big?
Nine days. People thought nine days of silence was worth asking about.
It turned out they were indeed cooking up something big—a complete overhaul of permissions and ecosystem. But I'm more interested in the rhythm itself. Previously, when building software, planning in weekly cycles seemed normal. Now if an idea hasn't shown any shadow after a week, people basically stop paying attention.
First Bottleneck: Passion
At this speed, what actually blocks startup teams?
I think the first thing is passion. Not the motivational-speaker kind.
Changes are too fast, and attention shifts too quickly. Trying to persist in such a massive wave purely for profit is unrealistic. Machine production speed has increased, but demand hasn't kept up. People still make decisions slowly; consumers are too lazy to think or switch. This inertia is currently the biggest source of stickiness, while technology itself offers no real moat.
You have to find joy in the thing itself. Otherwise, you won't last long.
Second Bottleneck: Focus
Twenty percent of energy here, thirty percent there, forty percent in another direction. You could barely get away with this before. Not anymore.
The entire industry feels like a 100-meter sprint. If the core team is still pacing themselves for a marathon—or even a brisk walk—they'll soon find people passing them by in waves until they can't see them anymore.
Direction follows the founding team's level of commitment. Whether a small team can fully dedicate themselves to one thing—this focus directly determines whether they can keep up with the pace. Teams with divided attention won't survive long at this velocity.
Third Bottleneck: Infrastructure and Process
Some bottlenecks are surprisingly not at the technical level, but in very basic places.
The product is ready, you're preparing to launch, and you discover the domain needs to be filed for registration—5 to 20 business days. Overseas? Register the domain and it's usable immediately. Just this one step differs by an order of magnitude in speed.
Similar issues include internal organizational approval processes, spending velocity, and decision-making chains. Before, these were just a bit slower, nothing noticeable. But when technology has reached this magnitude of speed, the surrounding parts that can't keep up become glaring obstacles.
Fourth Bottleneck: Physical Health
This one surprises many people.
Everyone thought AI would make things easier. But currently, that's not the case. AI still can't operate autonomously 24/7 unsupervised, so human consumption has actually increased.
You can run five agents in parallel, building and testing simultaneously. But human attention has to switch rapidly between several tasks, constantly checking results and correcting direction.
Honestly, the mental energy consumed by directing a team of agents is greater than directing a team of humans. It's like doing high-intensity workshops all day long—not occasionally, but every single day.
If your body can't handle it, you'll fall behind quickly.
Transitional State
My feeling right now is that this is a transitional period.
If some core directions and value chains begin to be validated, you can build a 24/7 AI loop around them, so things no longer rely entirely on human energy. Plus, if you can attract more contributors to participate, it's no longer just one person or a small team pushing forward on tokens alone—more people's brainpower and resources can join in.
This is already happening with our own open-source projects. Someone submitted a feature entirely using an AI coding agent, and it got merged—and this person isn't a programmer. As long as you have ideas, you can participate; identity shifts from consumer to builder.
The people at the table next to me during lunch—anxious in the first half, excited in the second—these two states will probably persist for a long time. It's just that in this process, you have to know exactly where you're stuck.