everyone can demo an AI product. very few can make it reliable enough that people trust it at scale. this tuesday in SF, we're bringing together founders building conversational AI in production to talk about the messy stuff: on june 9, 5:30–8:00 pm PDT at @frontiertower , join founders building AI in production: @srimalireddi ( co-founder, @usemoss (YC W25), @danylo_dev ( co-founder @lopus_ai YC W25), and @IT_Kabootar ( Co-founder @ @maximem_ai ), moderated by @itsshashank ( co-founder @noveumai ) the convo will be around hallucinations, evals, observability, reliability, and the messy reality of shipping AI products people can actually trust apply for an invite: luma.com/Conversational-AI-A…

I’ll be the @twilio Signal 2026 in SF. Come say hi!!

Just received the first printed version of the book

I am hosting Vibe-Code a Voice AI Agent in 3 Hours — Live Workshop with Claude Code - Open Registration. Join me! luma.com/vibe-code-a-voice-a… via @LumaHQ

drop ur startup link

I somehow go blank when I need to write a tweet!

It was great to hear @sama and @patrickc at Stripe session yesterday.

"LLMs are just predicting the next token." Technically true. Practically useless. That's like saying your brain is "just firing neurons." In my book, I break down what's actually happening inside a Transformer — the attention patterns, the emergent capabilities, the reasoning chains. Way more interesting than "next token prediction." Get it here -> intelligenceatscale.ai

The 3 things most AI tutorials don't teach you: How to evaluate if your LLM is actually working (not just vibes) How to handle failures in production agent systems Why does your RAG pipeline returns garbage when the context window gets too long I cover all three in my book "Intelligence at Scale" intelligenceatscale.ai

What part of the AI stack do you feel like you're still pretending to understand? I tried to answer all of it in one book. intelligenceatscale.ai — pre-orders open from $4.99

First 3 chapters of my book are free. No card. No signup. Just drop your email and get them in 60 seconds. If the writing doesn't hook you, the full book probably isn't for you. intelligenceatscale.ai ---

Everyone's building with AI. Almost no one understands how it works. I wrote the book I wish existed when I started. intelligenceatscale.ai

Just got my pass for Stripe Sessions 2026—looking forward to meeting up in San Francisco in April! stripesessions.com

The hottest space in tech is also the worst-explained. I spent 18 years on this. I finally wrote it all down. 22 chapters. 65K words. 80 diagrams. $4.99 Early Bird Pricing -> intelligenceatscale.ai

I wrote a book. "Intelligence at Scale" — LLMs, agents, and production AI. Everything I know from SageMaker, Prime Video, and building Noveum. Pre-orders open. $4.99 early bird → intelligenceatscale.ai

I am going to be the AI Impact Summit in Delhi with my cofounder and team. Come say Hi. @noveumai

AI will not replace engineers!! What's actually happening: → One engineer AI = 3-5x output → 3-month projects shipping in 5-7 days → The bottleneck of "how much code humans can write" is gone But here's the thing: We won't have fewer engineers. We'll have MORE software. More SaaS. More automation. More systems. Architects got CAD and simulation tools. Did we fire them? No. They started building bigger things. History repeating itself. Look forward to more awesome SAAS and software.

People think they can build a whole SAAS end to end using AI better internally than a company who has been building the same SAAS for years and now also has AI. 🤔

API.Market update: • 10K users globally • 6M API calls/month • 300 MCP servers • 30% MoM growth • $100K ARR No paid ads. Just building what developers need. The unsexy secret? Talk to users. Ship fast. Repeat.