One Useful Thing (Ethan Mollick)x.com · 19 September 2026

I think there are lots of good reasons to worry about whether AI is giving correct answers (especially free models), but, given the contradiction with other recent research suggesting AI is now doing well at financial advice, I was curious about what the methodology was here. I downloaded the report, which is from a company selling different AI financial services products. Given the limited information in the report, it is impossible to judge it's accuracy, since there are only limited examples of questions. However, I am curious if anyone familiar with UK tax law knows whether GPT-6 Pro's defense of Haiku's answer is right, or if Saturn's critique is.

Hiten Shahx.com · 19 September 2026

This is a wild amount of engineering to solve what is becoming a very important problem. How does the agent know the work actually worked? I suspect verification ends up eating a lot more of the agent stack than people expect.

Vantajobs.ashbyhq.com · 18 September 2026 · 2 jobs

At Vanta, our mission is to help businesses earn and prove trust. We believe that security should be monitored and verified continuously, and we empower companies to practice better security and prove it with ease. Vanta has a kind and talented team, and while some have prior security experience, many have been successful at Vanta without it. As a Senior Product Designer, Agent Experience at Vanta, you will define how AI agents help security and compliance teams accomplish meaningful work with greater clarity, control, and trust across Vanta and the surfaces where our customers already work. The Agent Experience team builds agent-led experiences that help security and compliance teams achieve outcomes with less manual effort. The team owns the agent harness, including agents, skills, tools, and memory, as well as the patterns that help customers understand, guide, review, correct, and trust agent-generated work. Our goal is to make AI-powered workflows useful, transparent, and dependable in a product where clarity and correctness matter. As a Senior Product Designer on this team, you will shape both the underlying agent experience and the c

LangChain bloglangchain.com · 17 September 2026 · 2 posts

Included Health is an all-in-one healthcare platform that partners with employers and health plans to provide their employees and members with healthcare navigation to services like virtual primary care, behavioral health, urgent care, specialty care, and more. The product experience centers answering medical, financial, or administrative questions via Dot—an AI-powered healthcare guide built on top of a federated multi-agent architecture using Deep Agents and LangGraph.The challenge: healthcare navigation doesn't fit a decision treeHealthcare is one of the few domains where what a person asks for and what they actually need can be entirely different. A member asking "is an artery plaque scan covered by my insurance?" might, with a few follow-up questions, reveal that they are managing elevated cholesterol and have a family history of heart disease. The right response includes the dollar figure—but it may also mean recognizing an opportunity to encourage a conversation with a primary care physician.Historically, health systems handled this kind of routing with structured navigation trees. That appr

YouTube: OpenAIyoutube.com · 16 September 2026 · 3 videos

Hey, I'm glad you're here. Today, we're going to do one simple thing together. Ask ChatGPT a question. You can think of ChatGPT like an assistant that never runs out of patience. You can type a question or say it out loud. If the answer is not quite right, you can ask another question. You don't have to get everything perfect the first time. Here's the message box. This is where you type. And that there, that's the voice button. Tap it to speak your message instead of typing it. Let's start with an easy question. Give me three easy ideas for using ChatGPT this week. And now, we get an answer. Maybe it suggests planning a meal, drafting a message, or learning something new. I don't have to use every idea. I can pick the one that sounds best to me. And now, I can keep the conversation going. Can you make that simpler and give me the first step? And that's really it. One question, then another. You don't need to do anything special. Just say what you want, look at the answer, and ask for it to be simpler, shorter, or more useful. Talk to it like it's a person. One important note. ChatGPT can make mistakes or have outdated informat