CEO: “we have thousands of AI agents running in production”
CTO:
25.07.2025 15:42
👍 1
🔁 1
💬 0
📌 0
CEO: “we have thousands of AI agents running in production”
CTO:
🚀 “Once you use AI, there’s no going back.” HIGH SIGNAL chat w/ @rauchg.blue, CEO of @vercel.com
1️⃣ V0’s explosion
2️⃣ “Vibe-coding” vs. agentic engineering
3️⃣ AI Cloud & Fluid Compute
4️⃣ MCP is the new HTTP
5️⃣ The 10-person GM model
6️⃣ Career advice for 18-year-olds
🎧 youtu.be/zhA7OMw4njw?...
AI coding is in full gold rush mode - excellent conversation with GitHub CEO Thomas Dohmke to understand vibe coding, the rise of coding agents, Cursor and the crazy pace of innovation in that space
youtu.be/Gp-oPGYrQDs?...
Doordash acquiring SevenRooms for $1.2B, an unprecedented $171M per room. We're so back.
🎧 Search, RAG and AI agents: My conversation with @arvindjain.bsky.social, CEO of Glean
When an entire administration suddenly realizes their kids are not going to get in
One of the secrets to success in a lot of leadership roles is the ability to deliver the same pitch thousands of time, with equal enthusiasm each single time.
My conversation with the CEO of Snowflake, on winning the race to become the dominant data and AI platform
Europeans, as US tariffs force them to rely exclusively on German cars, French wine and Italian fashion.
My conversation with star AI researcher @fchollet.bsky.social
and Zapier’s Mike Knoop
* ARC AGI 2 and why it breaks base LLMs
* test-time adaptation
* the search for fluid machine intelligence
116th edition (!) of #DataDrivenNYC Wednesday night with @matt-turck.bsky.social from @firstmark.bsky.social in conversation with CEOs from InfluxData, Sigma, OpenRouter, & Thread AI.
VCs when asked why they don’t just start a company themselves
This changes everything.
[insert model name] just rewrote the rules of what is possible.
Nobody understands the magnitude of change [insert model name] is about to bring.
Here are some INSANE examples of what you can do with [insert model name]. #10 is 🤯
It is so over.
AI in 2025:
O̶p̶e̶r̶a̶t̶o̶r̶ ̶i̶s̶ ̶h̶o̶t̶
̶D̶e̶e̶p̶ ̶R̶e̶s̶e̶a̶r̶c̶h̶ ̶i̶s̶ ̶h̶o̶t̶
̶D̶e̶e̶p̶S̶e̶e̶k̶ ̶i̶s̶ ̶h̶o̶t̶
̶G̶r̶o̶k̶ 3̶ ̶i̶s̶ ̶h̶o̶t̶
̶S̶o̶n̶n̶e̶t̶ ̶3̶.̶7̶ ̶i̶s̶ ̶h̶o̶t̶
̶G̶P̶T̶ ̶4̶.̶5̶ ̶i̶s̶ ̶(̶n̶o̶t̶)̶ ̶h̶o̶t̶
̶Q̶w̶e̶n̶ ̶i̶s̶ ̶h̶o̶t̶
̶M̶C̶P̶ ̶i̶s̶ ̶h̶o̶t̶
Manus is hot
If someone had told us just 2.5 years ago we'd have an omniscient, multimodal chatbot that can answer any question in the world, code, reason by analogy and show empathy and humor, most wouldn't have believed it, yet here we are, calling GPT 4.5 a disappointment lol
Give me your tired, your poor, your huddled masses and whoever has a spare $5M
"Don't do services" --> Palantir
"Hardware is hard" ---> NVIDIA, Apple, Tesla, Space X
"Selling to the government is awful" --> Anduril
If you want to build a true outlier, ignore generic startup advice.
Here's my conversation with the excellent @amasad.bsky.social, CEO Repliy. With the Replit AI Agent, the company has a big hit on its hands.
We chatted about the Replit journey, AI agents and the future of Silicon Valley.
youtu.be/9xhDL2GbzaU?...
I think about this almost every day
* Fund o3 mini-high: substance-enhanced decisions (our best performing)
* Fund Operator: we remind founders about our operating background (4 months at Salesforce)
* Fund Deep Research: we take meetings to research competitors of the companies we're actually interested in
Excited to announce a new series of funds:
* Fund 4o: good for most investments
* Fund 4o with scheduled tasks: we actually follow up
* Fund o1: we use reasoning (not just vibes) to decide whether to invest
* Fund o3: fast investment decisions (only 10 meetings and weeks of dd)
Ok folks, false alert: we checked, DeepSeek is not a real startup! The founder didn’t go to Stanford or Harvard, and they didn’t even do YC, or raise on a $100M cap SAFE for their pre-seed, so, nothing to worry about. Phew.
Microsoft: "we're good for our $80B"
Meta: "we're good for our $65B"
DeepSeek: "we're good for our $5.5M"
Fascinating that agent mode (a la OpenAI operator etc but UiPath back in the day) continues to mean replicating a human that clicks around on a website vs having machines talk machine to each other (API mode). Maybe that changes over time as agent-specific infra gets built.
VC after making one intro
Yes the book is really awesome 👍
This is truly a *must-watch* for anyone who wants to actually understand AI
@chiphuyen.bsky.social breaks down key concepts amazingly well in this discussion about how Generative AI is built and deployed
youtu.be/p7F4f42iZ-c?...
Links:
- Spotify: tinyurl.com/yc2kejhh
- Apple Podcasts: tinyurl.com/ywudhtv4
- YouTube: youtu.be/p7F4f42iZ-c
What you MUST know about AI engineering in 2025
Awesome episode with @chiphuyen.bsky.social
about her brand new book
All links in next post below:
Lol going from 111,207 followers on X to 71 on Bluesky is quite the humbling experience so far 😂
