🔥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

A+ headline, most Dutch thing I’ve ever read.

LLM agents are too expensive and too unreliable. Unfortunately, building agentic workflows that work beyond a good demo is hard. I talk daily to people who are trying, and it's tough. I have a paper and a few experiments to show you. A solution that cuts the costs of running an

A billion dollar valuation isn't cool. You know what's cool? A billion dollar seed round.

this is now! hop on

Hey im the CEO look at this cool prototype I built with Claude it could totally replace salesforce go do your thing and productise it asap ive already cancelled our contract

Move over "incorrect number of fingers", it's time for "improbable half-peeled bananas"

Not adding an “are you sure?” dialog for the “end meeting for all” action is such an egregious UX fail by Zoom Brought to you by me ending today's webinar halfway through 🙈

Imagine getting a chance to learn how the best companies in the world have integrated and scaled AI to billions of users. A hands-on session with somebody from Microsoft, GitLab, or Mastercard could teach you more than a year of self-prescribed YouTube videos. If you are in the

Posting from my drafts: Unfortunately this podcast is mostly wrong and I feel compelled to speak up because this type of AI twitter marketing is setting the field back by a few years. There are two types of problems addressable by AI: soft-decision boundary problems and

If I weren’t doing startups I’d put my energy into becoming this guy aeon.co/videos/how-res…

OK this is happening...

admire the boldness of a blog on 'robust generative AI agents' w/no actual solutions to make these robust Lots of monitoring & observability to quantify how flaky your system is, but still just a "prompt and pray" approach to having LLMs execute tasks. aws.amazon.com/blogs/machine-…

Let’s go! #calmsummit24

On the a16z pod martin_casado and I riff on how chatbots evolved from word2vec to gpt-4. We touch on agents, RAG, function calling, reliability, and briefly features me beatboxing.

Are LLMs a magic solution for enterprise customer support? martin_casado and Rasa cofounder and CEO Alan Nichol dive into the challenges and opportunities of AI-driven customer interactions. Here, they explore the hurdles of ensuring reliability, the sweet spot for LLMs

Such a great conversation with Alan Nichol about the limits of LLMs and on how to integrate a stochastic component (LLMs) into a more formal system (application).

MCP is all the rage, but for customer-facing AI "tool use" 🛠️ is actually the wrong abstraction. There's a new agentic pattern emerging which I've started calling "process use". It's a much better fit for voice and chat AI in customer support.