Wen Tian

Cofounder of SnapHop AI (

[Wen Tian @wensington]

The purpose of AI is to dissolve the application layer and reconnect human intent directly to computation. After forty years of accumulating interfaces, frameworks, and SaaS abstractions, the stack is beginning to unwind. At the bottom, what remains is compute, memory, storage, connectivity, and trust.

[Wen Tian @wensington]

@levelsio Agreed. If you need to pay someone to help you to use AI, then what's the point of AI?

[Wen Tian @wensington]

@levelsio Software is very easy to maintain if the only user is yourself.

[Wen Tian @wensington]

@dhh @XMoney

[Wen Tian @wensington]

@dhh @satyanadella Office apps such as Excel should be headless from now on. No need to learn how to do a pivot table using Excel UI anymore. My agent should just handle that.

[DHH @dhh]

Omarchy Quattro will be shipping this week.

[Wen Tian @wensington]

@rosyprosperity This is why I think DRAM/NAND needs will still go exponential from there: https://t.co/VLo4wyJowf

[Wen Tian @wensington]

@jvisserlabs As most human workflows become agentic, compute requirements will go exponential from here. We are only at the starting point. Here is a preview of what's to come: https://t.co/VLo4wyJowf

[DHH @dhh]

For anyone with endless ideas, this agent age is nirvana as those ideas are met with endless execution, endless exploration. I've never had has much fun working with computers as I do right now. What a time to be alive.

[Wen Tian @wensington]

@jukan05 Whatever happened to "I'm willing to go bankrupt rather than lose this race."?

[Wen Tian @wensington]

@HiCagr Everyone is a crypto trader now

[Wen Tian @wensington]

@HiCagr Does it really matter?

[Wen Tian @wensington]

@Chartfest1 1993

[Wen Tian @wensington]

@buccocapital Waiting for SaaS companies to trade 2x FCF before buying

[Wen Tian @wensington]

Most people's interactions with AI are still stuck in the 2023 chatbot era. They are essentially using AI as an answering engine. IMHO, the true inflection point of AI will be the "Netscape" moment of AI (since people like to compare this period to the dot com era for some reason). Essentially some sort of harness will be the new web browser. Claude code, cowork and codex are good but they are more like AOL and CompuServe (I am starting to date myself). We are still waiting for the "Netscape" of agentic AI.

[Wen Tian @wensington]

@buccocapital Both can happen. Just different time frames.

[Wen Tian @wensington]

@rosyprosperity The best case scenario is probably maintaining the current price level. Supplies will be tight for many years to come.

[Wen Tian @wensington]

Personal computer preference is unique to each individual. Some people like Windows so they buy PC laptops; some people like MacOS so they buy MacBooks; I like Linux so I buy PC laptops. MacBooks' advantages are not as big as they once were a few years ago. For an average user, they are still the best overall option. But the gap is quickly closing: https://t.co/HWJNA9OhPW

[DHH @dhh]

I've been living with the new XPS 13 for a few weeks now. This has to be the best deal in computing! It's got my favorite keyboard on any laptop, great screen, solid 18A chip, and it costs just $699?! HOW? I keep looking for the catch.

[Wen Tian @wensington]

@wallstengine Wait until Anthropic gets 50% of global SaaS revenue.

[Wen Tian @wensington]

@jukan05 Also, hyperscalers should implement price bidding for frontier models during peak usage hours. High quality tokens are simply too valuable.

[Wen Tian @wensington]

@jukan05 They won't. If they cut capex, their share prices will fall even faster. Hyperscalers are like like airlines now. Airlines need more planes to generate more revenues. Hyperscalers need more compute to generate more revenues.