Everyone is talking about agent loops and loop engineering right now, but most of the advice assumes you are a hardcore coder running fleets of agents around the clock. In this video I break down what an agent loop actually is (reason, act, observe, repeat), why the verification step matters more than the architecture, and how to think about a "done" criteria that your agent can actually check. I walk through a few real loops I ran, including thumbnail scoring, a three.js plane, and an Abbey Road recreation, and explain why loops are about getting you closer on the first try, not perfect output. If you have been feeling behind because you are not running five agents at once, this one is for you.

Most people think the biggest mistake in AI is not using it. I disagree. The biggest mistake is using AI without changing the way you think and work. Many people ask AI to do more. Very few ask: What process should I redesign? What task should disappear completely? What decision can be improved with better information? AI is not just a tool for faster work. It's an opportunity to rethink how work gets done in the first place. So here's my question for the community: 👉 What is one process in your work or life that should no longer exist in its current form because AI can do it better? Not "faster." Better. I'm curious to see what everyone comes up with. 👇

Hi guys I'm new to AI, where do I start? Please help me.

Hey everyone! 🙏 I could really use some guidance and I'd be super grateful for any tips. So here's my story: I'm pretty early in my AI journey, still learning a lot, but I went to a 3-day seminar, got talking to people, and mentioned I build AI automation and software. One of them turned out to be the CEO of a big distribution company. I built her a custom inbox-assistant demo for her business and she absolutely loved it. Now I've got a call coming up to talk price and how we move forward, and honestly I'm a bit nervous because it's my first real client and she's a CEO of a serious company. She knows I'm early in this, which helps, but I still want to do it properly. Her situation: she gets 200 to 300 emails a day in Outlook (Microsoft 365). She wants the AI to sort and prioritize her inbox, write drafts in her tone (English, Dutch, French), and give her a daily brief. The company is fully Microsoft, so it'll need their IT to approve the security side. I'd love any guidance on a few things: 1. Best way to connect securely to Outlook/Microsoft 365? Microsoft Graph directly, or a provider like Nylas/Unipile? Want it to pass their IT review. 2. How would you price something like this for a real company (setup + monthly)? 3. Anything to watch out for when working with a Microsoft company that has IT security checks? 4. Any tips for keeping the AI draft-only so nothing sends without her approval? I've built the demo and even put together a proposal, just want to make sure I deliver this well. Thank you so much in advance, this community is awesome! 🙏

Hey everyone, I'm currently building a voice agent using Vapi and Retell with Google Calendar integration. The agent is performing really well overall: - Speaks naturally and politely - Stays focused on the conversation - Answers questions correctly - Successfully books appointments on Google Calendar However, I'm running into an issue with rescheduling and cancellation. When I ask the agent to cancel or reschedule an appointment, it confidently responds with something like: "Your appointment has been cancelled." or "Your appointment has been rescheduled." But when I check Google Calendar, the original appointment is still there and nothing has actually changed. So my question for those of you who are actively using Vapi or Retell: - Do their built-in Google Calendar integrations fully support appointment cancellation and rescheduling? - Or is the integration mainly designed for booking new appointments? - Do I need to build additional automation/workflows using tools like n8n, Make, or custom API calls to properly handle cancellations and reschedules? I'd love to hear how others have solved this problem and what architecture you're using in production. Thanks in advance! 🙏