Loading...
Loading...
Loading...
# Demo Video Script (30-60 Seconds) ## π₯ Recording Setup ### Tools You Need: - **Loom** (free) - loom.com - **CapCut** (free) - for adding text - **Or Screen Studio** ($89) - if you want pro quality ### Browser Prep: 1. Open **Incognito/Private window** 2. **Hide bookmarks bar** (Cmd/Ctrl + Shift + B) 3. **Zoom to 90%** (Cmd/Ctrl + -) 4. **Close all other tabs** 5. **Full screen browser** (F11 on Windows, Cmd+Ctrl+F on Mac) ### Screen Resolution: - Set to **1920x1080** (HD) - Or **2560x1440** (better quality) --- ## π Script (45 Seconds Total) ### Scene 1: Landing Page (0:00-0:08) **What to show:** - Open http://localhost:3000 - Scroll slowly down the page - Pause on the form for 2 seconds **Text overlay:** ``` "AI-powered client intake" ``` **Cursor movement:** - Slow, deliberate - Highlight "Qualify leads while you sleep" --- ### Scene 2: Fill Form (0:08-0:15) **What to do:** - Click into name field - Type fast (or speed up 2x in editing): - Name: "Sarah Chen" - Email: "[email protected]" - Service: "Need help building a landing page for my SaaS" - Budget: "$3,000" - Click "Continue β" **Text overlay:** ``` "Client fills out form" ``` **Tip:** You can type slower and speed it up 2x in CapCut --- ### Scene 3: AI Conversation (0:15-0:30) **What to show:** - AI's first message appears - Type response: "We're launching in 2 months" - AI asks about budget - Type: "Around $3-5k" - AI asks one more question - Type answer - AI says: "Great news β you're a perfect fit. I'll send you a booking link shortly." **Text overlay:** ``` [0:15] "AI qualifies in real-time" [0:25] "Detects good fit automatically" ``` **Tip:** Keep responses SHORT. Don't actually have a long conversation. --- ### Scene 4: Email (0:30-0:38) **What to show:** - Switch to Gmail/email - Show the "Let's talk" email - Highlight the Cal.com booking link - Maybe hover over it **Text overlay:** ``` "Booking link sent instantly" ``` **Tip:** Have email open in another window, quick switch (Alt+Tab) --- ### Scene 5: Dashboard (0:38-0:45) **What to show:** - Switch to http://localhost:3000/dashboard - Show the stats (1 qualified lead) - Scroll to show the lead card - Click on it to show conversation history **Text overlay:** ``` "Everything tracked in dashboard" ``` **Ending frame:** ``` "Open source on GitHub github.com/yourname/repo" ``` --- ## π¬ Alternative: Speed-Run Version (30 Seconds) If 45 seconds is too long: ``` 0:00-0:05 Landing page (quick scroll) 0:05-0:10 Form fill (2x speed) 0:10-0:20 Chat (3 messages total, sped up) 0:20-0:25 Email arrives 0:25-0:30 Dashboard view + ending ``` --- ## βοΈ Editing in CapCut (Free) ### Steps: 1. **Import your Loom video** 2. **Speed up boring parts:** - Form filling β 1.5x speed - Typing β 2x speed 3. **Add text overlays:** - Use "Text" β "Default" - Font: Inter or Montserrat (built-in) - Position: Bottom third - White text, black background box 4. **Add zoom-ins (optional):** - When AI says "perfect fit" - When email arrives 5. **Export:** - 1080p HD - 60 FPS - MP4 format --- ## π± For X/Twitter ### Video Requirements: - **Max length:** 2 minutes 20 seconds - **Max size:** 512MB - **Recommended:** Under 1 minute for better engagement - **Format:** MP4, MOV - **Resolution:** 1920x1080 or 1280x720 ### Posting Tips: 1. **Upload directly to X** (not YouTube link) 2. **Add captions** - many watch muted 3. **First frame matters** - show something interesting 4. **Pin the tweet** on your profile ### Caption for X: ``` Built an AI intake agent that qualifies leads while you sleep β Instant 24/7 response β Smart qualification β Auto booking Built with Groq + Next.js + Supabase Completely open source Full demo π ``` --- ## π― Pro Tips ### Make It Feel Real: - Use a REAL name, not "Test User" - Use a REAL-looking email (not [email protected]) - Write REALISTIC project descriptions - Make the budget seem real ### Cursor Movements: - **Slow and deliberate** - not frantic - **Highlight important text** by hovering - **Pause on key moments** (when AI makes decision) ### Avoid: - β Mouse moving all over the place - β Typing errors that you backspace - β Switching windows frantically - β Background notifications popping up - β Messy browser with 50 tabs ### Do: - β Clean browser, one tab - β Smooth cursor movements - β Deliberate pacing - β Clear text overlays - β Show the value, not just features --- ## π¨ Optional: Add Branding If you want it to look EXTRA professional: 1. **Add intro card (2 seconds):** ``` Black screen with white text: "IntakeAI" "AI Client Qualification" ``` 2. **Add outro card (3 seconds):** ``` "Open source on GitHub" "github.com/yourname/repo" "β Star if useful" ``` --- ## β Checklist Before Recording - [ ] Close all browser tabs except the app - [ ] Hide bookmarks bar - [ ] Clear browser history/cookies (clean look) - [ ] Use incognito mode - [ ] Zoom browser to 90-100% - [ ] Close Slack, Discord, notifications - [ ] Test your microphone is MUTED (you don't need audio) - [ ] Have email open in another window for quick switch - [ ] Prepare the lead data in advance (name, email, etc.) - [ ] Run through once before recording --- ## π After Posting ### Track Engagement: - Views on first hour - Likes/RTs in first 24 hours - GitHub stars (if you link repo) - DMs asking questions ### Respond to Comments: - Answer technical questions - Share how you built it - Offer to help others deploy ### Follow-Up Content: 1. "How I built this in 2 days" thread 2. "3 things I learned building an AI agent" 3. "Adding [feature] based on feedback" --- ## π― Final Output **Your video should:** - Be 30-60 seconds - Show the FULL user journey - Look clean and professional - Have clear text overlays - End with CTA (GitHub link) **After editing, it should feel like:** - A product demo, not a screen recording - Intentional, not rushed - Clear value prop, not just features - Professional, not "I made this in my basement" --- **Ready to record? Follow this script shot-by-shot. You got this! π¬**
[](http://colab.research.google.com/github/rinongal/stylegan-nada/blob/main/stylegan_nada.ipynb)
get signed picture and voice authorisations from our parents
Parses a structured video script, extracts all `Narrator:` blocks, and synthesises them into a single MP3 using Azure OpenAI TTS.
<img src="https://img.shields.io/github/forks/artkulak/text2youtube.svg">