Our marketing team is just me and ~40 AI agents. I finally got around to putting them into an "org chart", and it's actually really cool to see! Plus, laying them out this way by sub-function (social media, blog, email, community, partners, etc) has given me a bunch of ideas of

Fake work is a people problem not a tool problem

Here’s the link:

Teaching my first class on @MavenHQ (really nice platform @gaganbiyani) It's going to be about how to build a team of AI agents like this one. Let me know if you're interested!

Write simpler prompts. Give better context / examples.

It is much easier to raise money with no traction than with a little bit of traction.

Are cluely and align the same company and this is just another way for them to get attention

RIP Claude 3.5 Sonnet. What a great model

Wow bending spoons just bought AOL (which still exists!!)

AI concept of the day: Tool Calling Here's a 30-second explanation: 1. Large Language Models (LLMs) take in a text input (Prompt) and give you back the text that is likeliest to follow that (Response). 2. In many cases, they can create a good response based on the knowledge

Overhyped: image generation Underhyped: basic extraction and classification

Throwback to 5 years ago when going 0-$10m in 19 months was fast. Now if it takes you 19 days, you might as well shut the company down.

I want to completely stop (a) typing, and (b) reading from screens. So I'm trying to use audio models as much as possible to both type for me and read to me. For the last few days, I've been working on getting automatic audio briefings of my emails, and I've got it working

AI costs can be counterintuitive. It turns out we’re spending more money recomputing embeddings on files users add to their knowledge bases than we are on our natural language assistant, even though the latter has 100x the engagement and user value.

Want to get an AI MBA for free? I just put together a speedrun curriculum that will teach you everything you need to know in 30 minutes (no technical experience required). Here’s what we’ll cover: 1. Large Language Models 2. Prompt Techniques 3. Context Windows 4. RAG 5.  Tool

people who are bad at LinkedIn mock it people who are good at LinkedIn get lots of customers

MCP is just a slow, unreliable, expensive API

It took us 2 painful years to get our first paying customer. If I had done these three things, it would have been 4x faster: 1) pick a known category 2) read the mom test 3) be better than the incumbent in one important way for one important user segment

Running a hot dog stand is better preparation for a startup than being a PM at Google

Does ARR mean annualized run rate or annual recurring revenue?

How to get engagement on social media: 1. Start with nuanced thing you believe 2. Remove all nuance 3. Make your position 10x as extreme 4. Present it as fact