Collab365 Team

I just spent 150 hours creating a 16-minute film.. Every frame, every sound, every voice - generated by AI. This was made with VEO 3.1, which launched yesterday. The coolest feature? You can now set start and end frames, making it exponentially easier to stitch scenes together seamlessly. Here's what I'm thinking about: Within 2 years (probably less), someone is going to create a full feature-length film with AI that Netflix, Amazon, or Apple buys. When that happens, the floodgates open. 20 years ago, who would have predicted we'd have the power to create entire films with AI? But here's the thing people miss: AI won't completely replace humans, but it will impact millions of us. We are where we are. Nothing will stop it now. "The One They Keep" (the 16-minute film) took 150 hours. I directed every decision. I wrote the narrative arc. I chose what worked and what didn't. I used AI for execution, script refinement, storyboards, animation, voiceover, sound design, music. The tool didn't make the film. I did. But I made it WITH AI. It supercharged me. And that's the divide forming right now: → People who can use AI as leverage → People who can't The gap between those two groups is widening every single day. The tools are democratized. The barrier isn't access anymore. It's willingness to learn. It's putting in the 150 hours. It's being comfortable being bad at something new until you're good. If you care about where we're heading as a society - and what it means for your career - I'd encourage you to watch "The One They Keep." It's on the Post Below.

🎬 "THE ONE THEY KEEP" I just got very carried away and created an entire short film completely with AI. Images, Research, Video, Sound Fx, Music, Script, the lot! BUT, It still took me 150 hours to make! In all seriousness, the film has a very important message that we should all start thinking about.

🆕 #KashBox article | If you like simple automations for your information and content, this article is for you. ⚙️ Microsoft Lists rules + demo ⚙️ @SharePoint quick steps + @WonderLaura vid ⚙️ @MSPowerAutomate flows + @Teachers_Tech vid Read & share 📄 kashbox.substack.com/p/automate-you…

‘This guy just automated his business in 15 minutes with AI.’ 👆I literally saw that on X yesterday and it annoyed the hell out of me. And honestly… it rattled me like many of these posts do. Not because I believed it. But because it’s everywhere right now and it causes us to feel anxious. Scroll for five minutes and it feels like every job on the planet is being automated. Like we’re all just waiting to be replaced by a prompt. And yeah - I worry about that. I worry about the future. About my kids. And about the people who’ve spent years building a solid, honest career… now wondering if it still matters. So, I did what I always do when things feel uncertain. I looked back through history. Bit of a Ray Dalio moment, if you know him. He studies how empires rise and fall — and what history can tell us about what’s coming. (Worth looking him up on YouTube). So I went into research mode. I asked one of those AI research tools: ‘Which jobs have actually been wiped out by technology? And which ones… just evolved?’ I looked at 12 jobs we’ve all heard of. And surprisingly… only one truly vanished. The rest? They shifted. They adapted. They changed shape — and the people behind them found new ways to stay relevant. That gave me a bit of peace. Because yes, AI will change the world. But it won’t run it. Not without oversight. Not without regulation. Healthcare. Legal systems. Aviation. Finance. Social services. Even the military. These sectors aren’t handing full control to AI anytime soon — if ever. So here’s my advice: If you work in tech — especially if you’ve ever automated anything — you’re probably going to be just fine. You already know how to pivot. But if your role’s mostly repetitive — and non-technical — now’s the time to start paying attention. Learn to prompt properly. Try no-code tools. Automate something just for fun. E.g. I built a workflow the other day that summarises YouTube videos and emails them to me. No code. Just curiosity. And it means I can learn ten times more, in a tenth of the time. It’s not rocket science — it’s just effort. So anyway… I just wanted to say this: If you’re someone feeling anxious about your future … Just know that if you put a bit of effort in now… You’re going to be just fine. ❤️

Anyone else feeling like AI is all going a bit too quick?

🧵 AI isn’t coming for truck drivers first. It’s coming for you. Most people think AI will replace low-skilled jobs first. But it’s already replacing: 💼 Admin assistants 📊 Junior analysts 🗂️ Project coordinators Why? Let me show you 🧵

Old job ad: “Looking for an organised assistant with great communication skills.” New job ad: “Looking for someone who can design automations and let AI handle the admin.” That shift is already underway.

In 5 years, we won’t ask: “What do you do for work?” We’ll ask: “What can you still do that AI hasn’t mastered yet?”

I figured out how to get 10x better results from Cursor vibe coding and it has nothing to do with better prompts and will cost you nothing more than a cheap Cursor Subscription. The secret? MCP wrappers that let AI test its own work. Here's EXACTLY how to do it: ❌ The Problem Everyone Faces You're "vibe coding" or human coding in Cursor. You write beautiful code. You run it. It breaks. You copy-paste error messages back to AI. It "fixes" the code. Still broken. This back-and-forth kills your flow and wastes hours. ✅ The MCP Testing Method I'm building Playlist (my AI video app) and got tired of the manual testing cycle. So I wrapped my existing API with MCP endpoints: 1️⃣ Create MCP Endpoints* - Create, Read, Update, Delete playlists ✅ - Create, Read, Update, Delete videos within playlists ✅ - Query playlist data ✅ - Run basic functionality tests ✅ 2️⃣ Connect Cursor to Your MCP Now when I need to change code, I tell Cursor: "Test this by creating a playlist, adding 3 videos, then querying the results" 3️⃣ AI Tests Its Own Work Cursor automatically: - Calls my MCP endpoints - Validates the responses - Catches errors before I do - Suggests fixes based on actual test results ✅ Why This Changes Everything Before MCP testing: ❌ Manual copy-paste debugging ❌ Broken code reaching production ❌ Hours lost on simple bugs ❌ AI guessing what went wrong After MCP testing: ✅ AI catches its own mistakes ✅ Real-time validation of changes ✅ Faster iteration cycles ✅ AI fixes based on actual data 👩🏼‍🏫 The Non-Technical Explanation Think of MCP like giving your AI assistant a direct phone line to your app. Instead of you being the messenger between AI and your app, they can talk directly. When AI changes something, it immediately calls your app to see if it still works. Like having a QA tester built into your coding assistant. 👀 Real Example with Playlist **Old way:** 1. I write playlist creation code (or AI helps) 2. Manually test in browser 3. Find bug, copy error message 4. Paste back to AI for fixes 5. Repeat 3-4 times **MCP way:** 1. I need to modify playlist creation logic 2. AI immediately tests via MCP: "Create playlist 'Test Mix'" 3. MCP returns success/error instantly 4. AI fixes any issues before I even see them The difference? 20 minutes vs 2 minutes. 🥤 Getting Started (Simple Steps) 1. Identify your core functions (for Playlist: create, read, update, delete) 2. Wrap them with MCP (think of it as creating a simple API your AI can call) 3. Connect Cursor to your MCP (add it to your Cursor config) 4. Start testing everything ("AI, change this code then test it works") You don't need to be super technical. If you have basic CRUD operations, AI can help you build the MCP wrapper. 🪄 The Real Magic This isn't just about testing. It's about AI that understands your actual product. When AI can directly interact with your existing codebase, it makes better decisions about: - Database structure changes - User flow improvements - Error handling - Performance optimizations It's like the difference between describing your car to a mechanic vs letting them drive it.

Everyone's panicking about Google killing websites. But they're asking the wrong question. It's not "Will this happen?" It's "How do I get ready before my competitors do?" Here's your 5-step bot-proofing action plan: 🧵 1/ Audit your "bot readiness" TODAY. Can an AI easily find your: • Pricing? • Services? • Contact info? • Key differentiators? It's suprising how many fail this basic test. Including mine before we fixed it. 2/ Start with structured data (the 2-week win). Add Schema [.org] markup to your site. Create clear FAQ sections. Structure your product info consistently. This isn't sexy work. But it makes you instantly "findable" by AI. While competitors stay invisible. 3/ Build your first agent (not a chatbot). This isn't for ChatGPT to find you yet. It's preparing for when AI CAN self-discover and connect to your systems. Start with YOUR website handling actions: 🤖 Schedule demos 💳 Collect payments 📋 Qualify leads 📊 Share pricing Tools: Voiceflow, Botpress, or ChatGPT custom GPTs. When Google's MCP framework goes live, you'll already have the infrastructure. While competitors scramble to build from scratch. 4/ Write for retrieval, not humans. Forget fluffy marketing speak. AI wants: • Clear facts • Specific numbers • Structured formats • Extractable expertise Your content becomes a database, not a brochure. 5/ Test with current AI tools. Ask ChatGPT about your business. Check what Perplexity says about your services. See if Google's AI Overviews mention you. Pro tip: Add Chatbase to your site (link in first comment). Once we installed it, we found TONS of gaps where AI wasn't answering correctly. We added FAQs to product pages and organized everything by speaker, topic, etc. Game changer. Find the gaps. Fix them. Repeat. The businesses starting now will own their categories when this fully hits. While everyone else scrambles to catch up. Don't bookmark this. Pick ONE step and ship it this week. What's your first move going to be?

Google is quietly killing your website. The traditional “10 blue links” are vanishing. Soon, your homepage won’t be for people—it’ll be for bots. Here’s what’s coming (and why it matters): 🧵 1/ Search is going agent-first. Google’s rolling out AI Overviews—answers straight in the results. No clicks. No visits. Just summaries. If no one’s visiting your site, what’s the point of having one? 2/ So what is the point of a website, then? It won’t be a place people browse. It’ll be a system bots interact with. In other words, your website becomes an API for LLMs. 3/ Google’s A2A and MCP frameworks flip everything. Instead of linking out, Google will call your agent directly. That agent will: 🤖 Schedule demos 🛒 Share product info 💳 Handle transactions ❓ Answer detailed queries Your “website” becomes a back-end service for conversations. 4/ This kills SEO as we know it. You’re not writing for people anymore—you’re writing for retrieval models. Forget keywords. Now it’s about: 🧠 Structured data ✅ Verified sources ⚙️ Agent logic 5/ Your next lead? It’ll come from a chatbot. Someone asks Google: “Can I book a demo with [your company]?” Your agent handles the whole thing. No forms. No pages. Just frictionless action. 6/ Here’s the real problem: most businesses aren’t ready. They’re still optimising for clicks, rankings, and funnels. But if 70–80% of traffic gets absorbed into AI summaries, that model’s gone. You’ll need new ways to be discovered—and to convert. 7/ The opportunity: build for the bots. Start thinking like a data provider, not a content publisher. Design for: 🔌 Clear APIs 🧩 Embedded agent behaviour 🔍 Verifiable info LLMs can trust Be the source, not the forgotten link. 8/ Now here’s the question no one’s asking… How’s Google going to monetise all this? If users don’t click ads anymore, where does the money come from? 9/ Google will take a cut of transactions. Soon, you’ll check out in the chatbot. 📞 Book a call 🛍️ Buy a product 💸 Make a payment Google handles it—and takes a fee. 10/And premium chat access will be paywalled. Want longer chats, better answers, or saved threads? You’ll pay for it—just like ChatGPT Plus. Google moves from an ad model to a paid-agent model. Genius! 11/ So what does that mean for you? You need to reimagine your site today. It’s no longer a destination. It’s a service endpoint—one that bots talk to on your behalf. — This shift is bigger than mobile. If you’re still writing meta descriptions, you’re already behind. The takeaway? Read up on A2A and MCP. We aren’t there yet, but it’s coming …

🚀 12 Practical Ways Microsoft Power Platform Can Transform Your Business 🧵👇 Are you still drowning in manual processes? Struggling with outdated systems? It’s time to level up your workflows and apps! Here are 13 incredible ways businesses are using Microsoft Power Platform to work smarter, not harder:

Call to Action – "Your Turn to Shine! 🌟" You’ve got the tools—now it’s time to act! ✅ Choose your site type (Team or Communication) ✅ Explore the LookBook for inspiration ✅ Build your first draft and gather feedback 🚀 Ready to learn more? Collab365 Academy has SharePoint workshops and live sessions to boost your skills. Let’s build a SharePoint intranet your team will love! #SharePoint #Microsoft365 #Productivity

Real-World Example – "Alex’s Success Story 🌟" Remember Alex? He started with the LookBook’s "Showcase" template and customized it: 🎨 Applied company branding 🖼️ Added Web Parts for announcements & galleries 📅 Launched in phases to gather feedback Within weeks, Alex’s SharePoint site became the go-to hub for company news and updates! 🙌

🚨 STOP! Your SharePoint Intranet Doesn’t Have to Look Boring! 🚨" Be honest: Does your SharePoint site look more like a file dumping ground than an engaging digital workspace? Good news: You don’t need to be a designer to create a stunning, functional intranet! Stick with me, and I’ll show you exactly how to build a SharePoint site your team will love. 🧵👇