the next big unlock in intelligence will come from orchestration more on this in a couple weeks

we were promised agi and instead we got people falling in love with text autocomplete

systems thinking is incredibly important to understand how to build complex agents, we've never been able to build self-adaptive software before

meta-prompting is fucking hard tbh

guys only want one thing and it’s disgusting

For anyone using Humanloop (as it’s shutting down), we’ve put together a migration guide to Latitude 👉latitude.so/blog/humanloop…

grifters now drop their playbooks openly on linkedin

Picking the right model for an agent isn’t “just use the biggest one.” It’s a tradeoff between capability, cost, and latency. Here’s the quick framework we use at Latitude 👇

sam altman doesn't want you to know this

the only thing i see when someone says 996

we’re bullish on multi-agent architectures, here’s why

I've been playing with Latte for a while and it's great. Allows me to build in 1 or 2 hours what would take me days. If you're building agents, you should give it a try. We truly admire cesar.wtf and Gerard Clos resilience and we think this is 🔥

Let’s see the features we're launching👀 1. Latte, our agent copilot Latte is able to build, edit and deploy agents from a single prompt. We built him using Latitude itself😎 This is just v1: soon Latte will handle evals & testing (we have some cute Latte stickers too)

btw we have the best swag in the industry