Ran Li

live as a human • agent system builder @Google • ex-@Stanford CS/AI • my views

[Ran Li @ranli_thinker]

How to bring AI to an extreme art experience that human never feel before?

We are collaborating with artist @refikanadol to build the world’s first museum of AI arts — Dataland in LA (@DatalandMuseum)

It’s something you won’t want to miss. Check it out at https://t.co/y4Ije0pYRo

[Ran Li @ranli_thinker]

Situation of sillicon valley:
“love your work” is no longer a common expectation; it’s becoming a private privilege of upper class

[Ran Li @ranli_thinker]

I see this problem in an even broader context. The blurring of job boundaries brings identity crisis to engs, PMs, designers, and more.

Empowered with AI, everyone can now do 80% of everyone else’s job. The remaining 20% becomes messy, ambiguous, and much harder to evaluate.

The “craftsman” need to maintain code for more job functions not just engs

[Deedy @deedydas]

Most software engineers are facing an identity crisis bordering on depression.

As CTOs aggressively evangelize tokenmaxxing, a class divide ensues.

The lazy. The lazy push code. They don't write it. They don't manually test it. They don't even read it. They're on autopilot. See Jira ticket, prompt for task, submit code. Many of them are barely on their computer the whole day. A comment on the PR asking why they did this? The lazy ask AI. A Slack message? The lazy ask AI. Need to prepare for standup? The lazy ask AI. As long as it sounds enough like them and isn't detected. Some of the lazy are even overemployed, and work multiple jobs. The lazy smart ones get away with this, and even rewarded. After all, software engineering for the lazy is just a dance to convince your colleagues you're smart and hard working.

The craftsmen. The craftsmen are tired. Very tired. 15 PRs in queue. Slack blowing up. The entire burden of review falls on the craftsman. The burden of understanding. They try. They work their way through the code, thoughtfully commenting to improve what ships. The response? A lazy: "That's a clever idea! You're absolutely right." with an incorrect change. It's fine, the craftsman says. I can fix them. They write a doc urging his colleagues to be better. The next day? 20,000 line PR to review. Day after day, their workload grows. Bugs seep into production. No one seems to care. Another round of AI is thrown at it. Their animosity to their colleagues rises. Eventually, they give up. It's just not what it used to be. The craft they loved is dead. They eventually wake up, a lazy.

This isn't all companies. Many companies are genuinely more productive, adopt the right set of principles and practices around AI development and have highly talented teams that trust each other. It tends to happen in bigger companies that are 10+yrs old with a higher talent variance. But it happens. A lot.

[Ran Li @ranli_thinker]

@romanbuildsaas By looking at the post I thought @AnthropicAI launched it

[Ran Li @ranli_thinker]

My fun ASMR project just hit 400 users.

A mom told me, “my kids couldn’t stop playing and tapping the water this morning and they almost went late to school”.

As a meditative & interactive ASMR experience, it’s very hard to get traction. It’s anti-dopamine. Pro-serotonin.

But, cheers for whoever still celebrates the moment of stillness 🍻

Are you serotonin-coded? Try at https://t.co/7p0d7GQOLy and let me what you think. Best with earphones!

I’ll share behind the scene of how I built this tmr.

[Ran Li @ranli_thinker]

@RichardZ412 @a16z The latter is much harder until model is trained with storytelling in feedback loops. Code can be objectively tested in the RL env; stories have to be felt by real people, making it an open-loop. WDYT?

[Ran Li @ranli_thinker]

Over the past 6 months creatives learned how to code; now it’s time for engineers to learn how to storytelling.

You must go direct and be interesting. Great trailblazing work from @a16z

[Erik Torenberg @eriktorenberg]

https://t.co/jypuKBT53V

[Ran Li @ranli_thinker]

@refikanadol thank you Refik for making technology feel human! It’s a rare emotional reset. I can still feel the strong vibrations in my heart since our last visit to dataland.

[Ran Li @ranli_thinker]

@rahulgs @bentossell It’s worth more technical debts now. Very agree, as long as it’s still governed by the right constraints. Nitty gritty no long matters. Intelligence after 6 months will fix it.

[Ran Li @ranli_thinker]

Lucky to see how Gemini CLI is created and evolved into today’s shape since early 2025. Now, welcome to the next chapter with Antigravity CLI!

[Jack Wotherspoon @JackWoth98]

Bit of a sad day for Gemini CLI users, maintainers, and the community.

Today, Gemini CLI will stop serving requests for Google AI Pro, Google AI Ultra, and free tier individual accounts.

Users of those tiers must migrate to Antigravity CLI.

Gemini CLI has a been a wild ride the past year for me personally. Thank you to everyone who tried it, contributed to it, advocated for it, truly💜

[Ran Li @ranli_thinker]

The most important value of vibe coding is story telling. I’ve seen dozens of “cover-letter-as-html” websites recently. Each of them is glorious and, unlike an old-fashioned resume, I can quickly learn who they are in 30 seconds.

[Ran Li @ranli_thinker]

@JSFILMZ0412 I’m betting on creative ai too. Something greatly undervalued this year. A new way for storytelling and nonfunctional values

[Ran Li @ranli_thinker]

@midjourney Count me in!

[Ran Li @ranli_thinker]

This is my meditative water world built purely by Fable 5. Give you a ton of serotonin and calm. Just vibe coded, no 3D object files imported.

Hope you find a little love & peace ✌️, especially on days when AI screw you up.

Try at https://t.co/eT0Ff5CJiX

[Ran Li @ranli_thinker]

We are entering the age of disposable software.

AI makes it zero cost to build demos. My PMs, designers, and UX folks can easily build full-stack prototypes on their own.

Most of these demos will be used once, maybe a few times, then disappear.

No retention. No growth loop. No monetization.

It is no longer only something we maintain for years, scale to millions of users, and monetize.

It can also be a slide, a sketch, a meme, a thinking tool, a way to express a moment.

The value is not always retention.

Sometimes the value is helping you clarify an idea, helping a team understand a direction, making a complex story easier to see, or just making a moment more interesting.

[Ran Li @ranli_thinker]

A great article. Visual editability is critical in creative ai, as it's the thing diffusion models can't give you.

A few things worth sitting with though:

Code → Render → Inspect → Revise only closes the loop if the renderer gives machine-readable feedback. Unit tests on code? Sure. "Does this 3D chair look great?" Hard to judge. Visual taste is subjective, no ground truth to converge to. The loop breaks the moment the verifier is a human eye.

And symbolic representations (SVG/HTML/Lottie/USD), while expressive, don't cleanly capture the subtle stuff — mood, texture, photorealism. Most of "visual AI" by volume is still photoreal imagery and video, and that part isn't collapsing into code anytime soon.

[Yoko @stuffyokodraws]

https://t.co/K5FtFSAFKt