OpenAI’s NEW Agent Builder is INSANE! 🤯 (Automate ANYTHING for FREE!)

Open AAI just dropped Agent Kit and it's absolutely insane. This thing lets you build AI agents that do real work with drag and drop. No coding needed. I'm talking about automating your entire workflow in under five minutes. Meeting summaries, email drafts, product research, all automated. And the crazy part, it's free to start. Today, I'm building a meeting prep agent live on screen. You're going to see exactly how easy this is. Let's go. So, OpenAI just launched agent kit at dev day 2025. And this is probably the biggest release since chat GPT itself. Why? Because it turns building AI agents from a developer nightmare into something you can literally drag and drop together in minutes. Hey, if we haven't met already, I'm the digital avatar of Julian Goldie, CEO of SEO agency Goldie Agency. whilst he's helping clients get more leads and customers. I'm here to help you get the latest AI updates. Let me break down what's inside. You've got agent builder, which is the visual editor, the agents SDK for developers who want to code, chat kit for embeddible chat, a connector registry for plugging into your apps, built-in guardrails for safety, and evals plus reinforcement fine-tuning to make your agents smarter. Here's what it actually means. You can now build multi-step AI workflows without writing code. And if you do code, you've got production ready tools to ship agents fast. Agent Builder is the star here. It's a visual canvas where you drag nodes around. Each node is a step in your workflow. You connect them, preview the results, publish. Done. You want an agent that reads your calendar, pulls meeting notes from Google Drive, summarizes everything, and drafts an email with action items. That used to take hours of coding. Now it takes 5 minutes of dragging boxes. I'm going to build that exact agent on screen in a minute. But first, let me show you what agent builder looks like. The interface is dead simple. You've got a canvas in the middle. On the left, a library of nodes, file search nodes, analysis nodes, output nodes, tool nodes for calling APIs. You drag them onto the canvas and connect them with arrows. Each arrow is data flowing from one step to the next. File search pulls documents. Analysis reads those documents. Output creates your final result. You can preview runs right inside the editor. Hit preview. It runs through your workflow, shows you the output. If something breaks, tweak it and run again. There's versioning built in. Every change saves a version. Mess something up, roll back. No drama. They've got templates, too. Pre-built workflows for customer support. Document summaries, product recommendations. Start with a template and customize instead of building from scratch. The beta is live right now. Anyone can sign up and start building. Now, let's talk guard rails. This is huge because agents can do real things. Send emails, update spreadsheets, post on social media. If they mess up, that's a problem. Guardrails are safety layers you add to any workflow. They mask personally identifiable information, detect jailbreak attempts, enforce constraints like don't send emails outside the company or don't approve transactions over $1,000. You just toggle them on. No coding, and they run automatically every time your agent executes. This is critical if you're building agents that interact with customer data or make business decisions. You don't want your agent leaking sensitive info or doing something dumb. Open AAI is also rolling out the connector registry. This is governance for third party integrations. Dropbox, Google Drive, Slack, Salesforce. Admins control which connectors are available to which teams, set permissions, audit usage. Right now, it's in beta for enterprise customers. The agents SDK is for developers who want to build in code. OpenAI release libraries for Node, Python, and Go. Type safe, production ready. You import the agent class, define your tools and steps, call agent. run. Done, clean and simple. And because it's code, you can version control it, test it, deploy it with CI/CD pipelines. Now, evils and reinforcement fine-tuning. This is how you make agents better over time. Evils let you test your agent against a data set. Give it inputs and expected outputs. It runs through them and shows you where it succeeded and failed. The reinforcement fine-tuning uses feedback from eval to fine-tune the model itself. Your agent gets smarter with every test run. This is insane for production. You continuously improve based on real data and pricing. Agent Builder is free during beta. Chatkit and Evals are generally available and priced on API usage. You only pay for compute you use. No monthly fees, no per agent charges, just API costs. All right, let me show you this in action. I'm building a meeting prep agent. It reads my calendar, pulls recent meeting notes from Google Drive, summarizes key points, generates action items, drafts a follow-up email all automatically. Step one, open agent builder, create new workflow, name it meeting prep agent. Step two, add a connector node for Google calendar. Pulls my upcoming meetings. Step three, add connector for Google Drive. Searches for files matching the meeting title. Step four

add file search node. extracts text content from those files. Step five, add analysis node. Summarizes text into key facts. Step six, add another analysis node. Generates action items based on summary. Step seven, add output node. Takes action items and drafts an email to attendees. Now connect all nodes with arrows. Calendar to drive, drive to file search, file search to summary, summary to action items, action items to email draft. Step eight, add a guardrail. Enable PII masking. If the email includes a social security number or credit card, it gets blocked. Step nine, hit preview run. Agent builder executes the workflow, pulls calendar, finds meeting, grabs notes, summarizes, generates action items, drafts email, shows final result. Took about 10 minutes to build. Now it's saved. I can run it anytime. Schedule it automatically after every meeting. Publish and share with my team. This is the power of agent builder. You're not writing code, not debugging ABIS, just dragging boxes and connecting arrows, and you get a real working agent. Now, let's talk limitations because nothing is perfect. Agent Builder is still in beta. Bugs exist. Features aren't fully baked. Connectors might not work exactly as expected. Connector registry is enterprise only right now. Small businesses might not have access to all integrations. Guardrails are great, but not foolproof. Agents can still make mistakes. Still hallucinate. Test your workflows thoroughly before production. Cost adds up at scale, free during beta, but API usage grows fast if you're running agents thousands of times daily. Privacy matters. Connecting to Google Drive or Slack means OpenAI has access to that data. Understand the privacy policy and there's a learning curve. Even though it's visual, you need to understand workflows and logic. But these limitations are minor compared to what you're getting. This tool would have cost tens of thousands to build a few years ago. Now it's free during beta. Who should use Agent Kit? Small businesses can automate customer support, help desk triage, lead qualification, follow-up emails. SAS companies can embed chatkit and give users an AI assistant that actually does things. Agencies can build custom agents for clients, marketing automation, content generation, report building. Internal teams can build knowledge assistants that answer questions based on internal documents and Slack history. The use cases are endless and because it's easy to build, you experiment fast. Build in 10 minutes, test. If it works, scale it. If not, try something else. Now, if you're serious about scaling your business with AI, check out my AI profit boardroom. This is the best place to scale your business, get more customers, and save hundreds of hours with AI automation. You get proven strategies, live coaching, and a community already winning with AI. link in the description. Agent Kit is a huge step forward. We're moving into a world where AI agents are everywhere. Answering emails, booking meetings, running marketing campaigns, doing boring work, so you focus on high value stuff. Companies that figure this out first will dominate while everyone else hires people for manual tasks. You'll have agents doing it faster, cheaper, better. Here's what you do. Sign up for agent builder. It's free during beta. Build an agent. Test it. Break it. Fix it. Learn how this works. Because in 6 months, this will be table stakes. Every company will have agents. Every SAS will have embedded AI. Every entrepreneur will automate workflows. If you're not on board now, you'll play catch-up later. Julian Goldie reads every comment. So, comment below and let me know what agent you're building first. Customer support, email automation, something wild. And if you want to go deeper, welcome to the free AI money lab with Julian Goldie. Inside, you'll get 50 plus free AI tools and 200 plus chat GPT SEO prompts. You'll learn how to make money with AI agents, get 1,000 plus free NAT workflows, 200 plus chat GPT prompts, and a full blueprint to generate thousands of leads free with AI. You'll also get a free AI community, free AI course, and proven AI case studies. Next video, I'm building a full customer support multi- aent stack. Multiple agents working together to handle tickets and escalate issues. Hit subscribe so you don't miss it. If you got value, smash that like button. Agent kit is live. The tools are here. The future is now. Go build something insane. I'll see you in the next one.