Want to know how AI will affect your industry? Look at hospitality.
It's the bellwether for how the technology actually shows up.
@GopiKallayil, Chief Business Strategist for AI at Google, breaks down 4 dimensions where AI is completely reshaping hospitality, and argues it's the best proxy for what's coming in every industry:
1) The customer journey
2) Marketing
3) Service delivery
4) Operations
The pattern holds across healthcare, finance, retail, and beyond. Hospitality just gets hit first.
Check out the full article below.
Everyone's running the same frontier models now.
What separates the great teams from everyone else is the operating system wrapped around the model.
The harness.
The interesting part isn't that harnesses exist... It's that every major AI provider has made a very different bet about what a harness should be.
New from @janakiramm: Four Harnesses, Four Philosophies 👇
Sam Altman is easy to read. He wants money, power, control, and the largest seat at the table. There's nothing mysterious or unique about that.
Dario Amodei's danger is subtler. He can explain, in morally serious language, why Anthropic acquiring power is the responsible thing to do.
A new piece from @DaveShapi 👇
"So many people discouraged me from working on the browser. But can you make this decision from offense, not just defense?"
9 months ago we interviewed @AravSrinivas CEO @perplexity_ai and it's amazing how much of what he said has already become reality.
"Offense is when there are things you can only do on the browser that you cannot do anywhere else. And that's where AI and search are headed next... agents."
"The first real agent everybody shipped was deep research. It can go research the web, do things. Then we built Labs, which can actually build dashboards, websites, analytics, web apps."
"Imagine that power thrown at daily browsing tasks. Deep research over your Slack, Notion, Google Docs. Answering your 100 emails. Auditing your calendar. All the stuff a personal assistant would do for you."
Introducing Claude Opus 4.7, our most capable Opus model yet.
It handles long-running tasks with more rigor, follows instructions more precisely, and verifies its own outputs before reporting back.
You can hand off your hardest work with less supervision.
“Software is dead.”
Joe Floyd on why the transition to AI is already over:
“I wouldn��t even call it a transition. The transition already happened. We have to burn the boats and fully commit to this AI future.”
“That means expanding our scope to things we never would’ve backed before, like hardware, infrastructure, and foundation models for robotics.”
“Ten years ago, that wasn’t Emergence’s sweet spot. Today, it has to be.”
“The traditional software model is simply dead. You can’t hire hundreds of sales reps, sell SaaS, and grow 40% a year for a decade anymore.”
“That playbook doesn’t work in this era.”
As AI content takes over the internet, what's the future of web browsing?
Aravind Srinivas, CEO of Perplexity says "Whenever I post a tweet on X, a ton of responses are bots. I keep marking them as spam. And it seems pretty hard to fight."
"There's a decoupling between the human and the actual internet. Now there's going to be an agent in the middle."
"Go read this article for me. Filter all the ones that look like AI and spam. Pull up all the signal, summarize it, give it to me in the format I want to ingest."
“A world without robots would be worse.”
@rdn_nikita CEO & Co-founder @FlexionRobotics says:
“Right now, AI is automating creative, academic, and intellectual work."
"But not manual labor.”
“So what happens? You ask AI how to fix your bike. It gives you step-by-step instructions. And you become the hands of the AI.”
“I think that world is much worse… Than one where we tell robots what to do. And they handle the manual work.”
7 of the top 10 companies by market cap depend almost entirely on TSMC's ability to produce chips.
Trillions in revenue. Concentrated in one specific cluster of fabs within a square mile in Taiwan.
That's not a supply chain, that's a single point of failure.
Perplexity started as a small business tool for ourselves. We had 4 people and no revenue with AI at our fingertips.
The pivot to Computer is actually a full circle. Founders are using it to grow companies that matter to the economy and their communities.
It’s rewarding to see it now powering small businesses and startups in big ways.
Perplexity is still a startup. We just 5X’ed revenue from $100M to $500M with only 34% growth in team size. 2x revenue growth in 2026 with same small team. And we’re just warming up.
Everyone here works at a small business, and everything we build is for people who build.
What’s better: robots with legs or robots with wheels?
@rdn_nikita CEO & Co-founder @FlexionRobotics says:
“A humanoid on legs can go anywhere a human can. Wheels are more efficient if you’re on perfectly flat ground and moving heavy loads over long distances.”
“But wheeled platforms hit limits very quickly. If the ground isn’t flat—steps, slopes, cables—wheels get stuck.”
“And there’s a second problem that’s even bigger.”
“With wheels, you either build a large, stable platform that doesn’t fit through doors, or you make it small and then it’s actually harder to control than legs.”
“Once you lift any weight, the whole system can tip over.”
“At that point, you lose the safety advantage that wheels were supposed to give you.”
"AI is not ready yet to completely do everything autonomously..."
@AravSrinivas CEO @Perplexity_ai on why the browser still matters:
"I still wouldn't trust Comet to accurately do financial accounting for Perplexity. We have a lot of cash and I wouldn't trust it to log into Morgan Stanley or JPMorgan — those are pretty hard systems."
"You want one environment where you do all your work. That's what the browser enables for you."
"The omni box is where you're typing most of your stuff. If I get to help you directly there, or on any web page you're on, I get to be on your side to co-browse with you."
"My token usage is probably down 30x compared to before.”
Tim Davis, Co-Founder & President @modular on why he builds AI tooling locally first:
“Token usage is incredibly expensive. If you just throw your whole codebase into a model, you’re constantly uploading everything and burning hundreds of thousands or millions of tokens.”
“Instead, you can build local representations of your codebase using embedding models. They’re more context-aware, and you only need to send much smaller summaries to models like Claude, Codex, or Gemini.”
“That gives you enormous efficiency gains. My token usage is probably down 30x compared to before.”
“It’s not going to scale for everyone to run their own agents.”
@zachlloydtweets on why agents need team-level visibility:
“I have agents running right now as we’re talking. The big thing we’re building is visibility and tracking into what they’re doing across a team.”
“We’re trying to make it so you can turn any skill into an agent. Skills are a really powerful primitive for this.”
“With our web app, you can see all the agents running across your company. You can search them, filter by PRs, and review their conversations.”
“It doesn’t scale for every developer to run four different agents on their own.”
“Companies need a centralized system where you can understand what all your agents are doing.”