🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

We've been thinking a lot about AI-powered work, but... what happens when AI systems don't work? Turns out that 👩🏽‍💻s still matter, and we can't ignore that in our products: open.substack.com/pub/frontierai…

.RunLLM now generates a weekly summary of its activity for you — including two new features that identify FAQs and flag uncommon questions!

Cursor and other code gen tools do not: 1. Help you understand the customer’s problem 2. Give you the right to co-create with customers 3. Give you any ability to innovate architecturally Will help good teams move faster, but does not solve the value creating R&D vectors

.RunLLM has been supporting the Phoenix community with Arize AI for about month now, and it's been really neat to see what people are building with Phoenix! Thanks to the Arize team for all their feedback. 🙂

The weird stuff really never stops!

AI systems are going to have to learn how to use modern SaaS in the same way that people do — which means knowing when to ignore your Slack messages! Joey Gonzalez & I on why building integrations for AI is harder than we realized: open.substack.com/pub/frontierai…

AI products not offering try before you buy is baffling to me. I am obviously pro AI, but it feels way too early for companies to not be trying to build trust!

Back in 2023, my students working on Gorilla project made the case for connecting LLMs to SaaS APIs. Today, everyone knows that models should be interacting with APIs. What people don't realize is that LLMs need to know more than how to call the API. They need to learn the

Feels like new phone camera features have diminishing value. No one has known what the camera specs mean for a few years now, but the editing/AI features are losing most people now too. 😬 The AI features have to deliver, or no one's going to upgrade at all.

A new kind of spam text I've gotten a ton of recently: fake recruiters. 🤦‍♂️ Except they're not recruiting for any well-known companies — the names sound mostly made up.

.RunLLM is now live on the Juicedata community docs — check it out!

.Joey Gonzalez and I have concluded we need to use more AI tools — both because it makes us more productive but also because it helps us understand how to build our own products better. Here's why we think that applies to you too: open.substack.com/pub/frontierai…

o1 feels like the most controversial release in a while, and (not coincidentally?) it also feels like a relatively high-variance model. Sometimes you get great reasoning and sometimes you get weird stuff. Kinda like humans!

Football numbers being flexible is generally good, but kickers with numbers in the 30s and 40s somehow just seem ridiculous.

.RunLLM's live on the MLflow docs site! Super excited to be supporting one of the biggest & more mature ML communities out there. Thanks to the Databricks team for the quick feedback & turnaround.

MLflow docs now have an AI assistant powered by RunLLM! The AI chat widget helps you: 🔍 Find info fast 💻 Get code examples 🧩 Understand MLflow components Click "Ask AI" on any docs page to get started! #mlflow #ai #LLM #MachineLearning

How long until people prompt typos into LLM outputs to make them seem more human? 🤔

.RunLLM's now live on the TorchMetrics docs from Lightning AI ⚡️ — check it out!

.Joey Gonzalez and I have been thinking about whether we should anthropomorphize RunLLM. We don't have an answer yet, but it's a hot topic, so we thought we'd share our thinking: open.substack.com/pub/frontierai…