This Indian engineer working at Netflix just built an open source tool that cuts your AI token usage by 60 to 95%.

{
  "hook": {
    "spoken": "This Indian engineer working at Netflix just built an open source tool that cuts your AI token usage by 60 to 95%.",
    "ends_at": "00:05.8",
    "on_screen": "this indian engineer",
    "starts_at": "00:00.0",
    "technique": "bold_claim",
    "technique_note": "Pairing authority (Netflix engineer) with massive metrics (60-95% token savings).",
    "why_it_stops_the_scroll": "Appeals directly to AI developers burning expensive API tokens with an immediate high-ROI solution."
  },
  "tags": {
    "topics": [
      "ai",
      "aitools",
      "developer",
      "opensource",
      "coding"
    ],
    "cta_type": "none",
    "language": "en",
    "has_music": true,
    "has_speech": true,
    "content_type": "showcase",
    "products_or_apps": [
      "headroom",
      "netflix",
      "github",
      "claude code",
      "cursor",
      "copilot"
    ],
    "has_burned_captions": true
  },
  "audio": {
    "music": "Subtle low-volume upbeat electronic synth background track",
    "voice": "Clear, studio-quality male voice with warm compression and crisp high end",
    "ambient": "Quiet treated room tone",
    "mix_notes": "Voice dominates mix heavily centered; background music remains barely perceptible",
    "sound_effects": "Subtle UI pop and swoosh on cut transitions"
  },
  "shots": [
    {
      "end": "00:01.0",
      "index": 1,
      "start": "00:00.0",
      "camera": "Static eye-level",
      "framing": "Medium close-up host bottom, photo top",
      "movement": "Static",
      "shot_type": "split_screen",
      "transition_in": "Cut",
      "screen_content": "Outdoor photo of engineer with ocean backdrop",
      "subject_action": "Gestures and talks to camera",
      "what_is_visible": "Host talking at bottom; photo of Tejas Chopra outdoors on top",
      "duration_seconds": 1,
      "spoken_during_shot": "This Indian engineer",
      "on_screen_text_during_shot": [
        "this indian engineer"
      ],
      "facial_expression_or_gesture": "Open hand gestures, engaging facial expression"
    },
    {
      "end": "00:02.1",
      "index": 2,
      "start": "00:01.0",
      "camera": "Digital animation",
      "framing": "Centered",
      "movement": "Motion graphic",
      "shot_type": "graphic",
      "transition_in": "Cut",
      "screen_content": "Animated Netflix logo",
      "subject_action": "None",
      "what_is_visible": "Netflix 3D red ribbon logo unfolding on dark background",
      "duration_seconds": 1.1,
      "spoken_during_shot": "working at Netflix",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "None"
    },
    {
      "end": "00:03.1",
      "index": 3,
      "start": "00:02.1",
      "camera": "Digital graphic",
      "framing": "Centered card",
      "movement": "Static",
      "shot_type": "screen_capture",
      "transition_in": "Cut",
      "screen_content": "GitHub UI showing repo file list",
      "subject_action": "None",
      "what_is_visible": "GitHub repo card for headroom repository",
      "duration_seconds": 1,
      "spoken_during_shot": "just built an open source tool",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "None"
    },
    {
      "end": "00:05.8",
      "index": 4,
      "start": "00:03.1",
      "camera": "Digital graphic",
      "framing": "Centered vertical card",
      "movement": "Animated bar chart",
      "shot_type": "graphic",
      "transition_in": "Cut",
      "screen_content": "Table with columns Workload, Before, After, Savings and growing percentage bar",
      "subject_action": "None",
      "what_is_visible": "Token usage table and comparative bar chart expanding upward",
      "duration_seconds": 2.7,
      "spoken_during_shot": "that cuts your AI token usage by 60 to 95%.",
      "on_screen_text_during_shot": [
        "Token usage",
        "60%",
        "95%"
      ],
      "facial_expression_or_gesture": "None"
    },
    {
      "end": "00:09.3",
      "index": 5,
      "start": "00:05.8",
      "camera": "Static eye-level",
      "framing": "Medium close-up host bottom, UI top",
      "movement": "Static",
      "shot_type": "split_screen",
      "transition_in": "Cut",
      "screen_content": "GitHub profile card followed by Headroom repo card",
      "subject_action": "Talking with two hands raised",
      "what_is_visible": "Host talking at bottom; GitHub profile card of Tejas Chopra switching to repo card top",
      "duration_seconds": 3.5,
      "spoken_during_shot": "His name is Tejas Chopra and the tool is called headroom.",
      "on_screen_text_during_shot": [
        "Tejas Chopra",
        "headroom"
      ],
      "facial_expression_or_gesture": "Informative and earnest expression"
    },
    {
      "end": "00:10.9",
      "index": 6,
      "start": "00:09.3",
      "camera": "Digital graphic",
      "framing": "Centered UI card",
      "movement": "Animated connection",
      "shot_type": "graphic",
      "transition_in": "Cut",
      "screen_content": "Diagram: App icon -> Headroom proxy -> AI model icon",
      "subject_action": "None",
      "what_is_visible": "Card showing Headroom in between Claude icon and AI circle icon",
      "duration_seconds": 1.6,
      "spoken_during_shot": "It sits between your app and the model.",
      "on_screen_text_during_shot": [
        "HEADROOM"
      ],
      "facial_expression_or_gesture": "None"
    },
    {
      "end": "00:12.7",
      "index": 7,
      "start": "00:10.9",
      "camera": "Static eye-level",
      "framing": "Medium shot",
      "movement": "Static",
      "shot_type": "talking_head",
      "transition_in": "Cut",
      "screen_content": "None",
      "subject_action": "Explaining concept with open palms",
      "what_is_visible": "Host at desk with microphone and laptop",
      "duration_seconds": 1.8,
      "spoken_during_shot": "So every tool output, log",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "Expressive explanation gestures"
    },
    {
      "end": "00:15.9",
      "index": 8,
      "start": "00:12.7",
      "camera": "Digital animation",
      "framing": "Centered hourglass graphic",
      "movement": "Animated particles falling through funnel",
      "shot_type": "graphic",
      "transition_in": "Cut",
      "screen_content": "Funnel graphic labeled Compressed with data tags filtering through",
      "subject_action": "None",
      "what_is_visible": "Logs and chat icons funneling down through a Headroom circular filter and shrinking",
      "duration_seconds": 3.2,
      "spoken_during_shot": "and chat history passes through it and gets compressed before it reaches the model.",
      "on_screen_text_during_shot": [
        "HEADROOM",
        "Compressed"
      ],
      "facial_expression_or_gesture": "None"
    },
    {
      "end": "00:18.9",
      "index": 9,
      "start": "00:15.9",
      "camera": "Digital graphic",
      "framing": "Centered vertical layout",
      "movement": "Subtle motion/beam",
      "shot_type": "graphic",
      "transition_in": "Cut",
      "screen_content": "Illustration of documents, folders, and sun/beam icon",
      "subject_action": "None",
      "what_is_visible": "Orange beam illuminating files and folder icons below",
      "duration_seconds": 3,
      "spoken_during_shot": "No summarizing and loss, just the same answer with way fewer tokens.",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "None"
    },
    {
      "end": "00:19.9",
      "index": 10,
      "start": "00:18.9",
      "camera": "Digital graphic",
      "framing": "Centered card",
      "movement": "Static",
      "shot_type": "screen_capture",
      "transition_in": "Cut",
      "screen_content": "Terminal banner graphic with Headroom text",
      "subject_action": "None",
      "what_is_visible": "Headroom GitHub README header card",
      "duration_seconds": 1,
      "spoken_during_shot": "Setup is simple.",
      "on_screen_text_during_shot": [
        "HEADROOM"
      ],
      "facial_expression_or_gesture": "None"
    },
    {
      "end": "00:21.0",
      "index": 11,
      "start": "00:19.9",
      "camera": "Static eye-level",
      "framing": "Medium close-up host bottom, terminal code top",
      "movement": "Static",
      "shot_type": "split_screen",
      "transition_in": "Cut",
      "screen_content": "Bash snippet showing pip install headroom-ai",
      "subject_action": "Speaking to camera",
      "what_is_visible": "Host talking bottom; bash pip install command snippet top",
      "duration_seconds": 1.1,
      "spoken_during_shot": "1 pip install,",
      "on_screen_text_during_shot": [
        "pip install \"headroom-ai[all]\""
      ],
      "facial_expression_or_gesture": "Calm, clear delivery"
    },
    {
      "end": "00:24.7",
      "index": 12,
      "start": "00:21.0",
      "camera": "Digital screencast",
      "framing": "Centered CLI window",
      "movement": "Typing and executing code",
      "shot_type": "screen_capture",
      "transition_in": "Cut",
      "screen_content": "Terminal screen running Headroom proxy with Claude agent",
      "subject_action": "None",
      "what_is_visible": "CLI executing headroom wrap claude command and running agents",
      "duration_seconds": 3.7,
      "spoken_during_shot": "then run headroom-wrap-claude or start it as a proxy with zero code changes",
      "on_screen_text_during_shot": [
        "headroom wrap claude",
        "HEADROOM WRAP: CLAUDE"
      ],
      "facial_expression_or_gesture": "None"
    },
    {
      "end": "00:26.7",
      "index": 13,
      "start": "00:24.7",
      "camera": "Digital graphic",
      "framing": "Centered dashboard card",
      "movement": "Zoom into savings counter",
      "shot_type": "graphic",
      "transition_in": "Cut",
      "screen_content": "Analytics panel displaying $25.55 savings, 5.3M tokens saved, 89% hit rate",
      "subject_action": "None",
      "what_is_visible": "Metrics dashboard showing compression savings dollar amounts and hit rates",
      "duration_seconds": 2,
      "spoken_during_shot": "and you instantly see the savings.",
      "on_screen_text_during_shot": [
        "$25.55",
        "5.3M",
        "89%"
      ],
      "facial_expression_or_gesture": "None"
    },
    {
      "end": "00:28.8",
      "index": 14,
      "start": "00:26.7",
      "camera": "Digital graphic",
      "framing": "Centered table",
      "movement": "Rows highlighting sequentially",
      "shot_type": "graphic",
      "transition_in": "Cut",
      "screen_content": "Agent compatibility table with green checkboxes",
      "subject_action": "None",
      "what_is_visible": "Agent support table highlighting Claude Code, Cursor, Copilot CLI",
      "duration_seconds": 2.1,
      "spoken_during_shot": "It works with Claude Code, Cursor and Copilot,",
      "on_screen_text_during_shot": [
        "Claude Code",
        "Cursor",
        "Copilot CLI"
      ],
      "facial_expression_or_gesture": "None"
    },
    {
      "end": "00:31.9",
      "index": 15,
      "start": "00:28.8",
      "camera": "Static eye-level",
      "framing": "Medium close-up host bottom, star history top",
      "movement": "Static",
      "shot_type": "split_screen",
      "transition_in": "Cut",
      "screen_content": "Star History line chart spiking to 30K stars",
      "subject_action": "Gestures with hands and concludes statement",
      "what_is_visible": "Host talking at bottom; Star History exponential chart at top",
      "duration_seconds": 3.1,
      "spoken_during_shot": "and crossed almost 30,000 stars on GitHub.",
      "on_screen_text_during_shot": [
        "30,000 stars",
        "Star History"
      ],
      "facial_expression_or_gesture": "Impressed, smiling expression"
    }
  ],
  "video": {
    "pacing": "Fast, informative tech-explainer pace",
    "setting": "Minimalist home office",
    "subject": "Tech creator speaking directly to camera with microphone and laptop",
    "lighting": "Soft key light from side, clean indoor ambient",
    "color_grade": "Natural contrast with subtle cool tones",
    "orientation": "vertical",
    "aspect_ratio": "9:16",
    "overall_format": "Split screen talking head mixed with motion graphics and UI overlays",
    "tone_and_energy": "Direct, enthusiastic, authoritative",
    "duration_seconds": 31.9,
    "wardrobe_and_look": "Dark crewneck t-shirt, short dark hair, full beard"
  },
  "script": [
    {
      "end": "00:05.8",
      "line": "This Indian engineer working at Netflix just built an open source tool that cuts your AI token usage by 60 to 95%.",
      "start": "00:00.0",
      "delivery": "High energy, punchy hook delivery",
      "uncertain": false
    },
    {
      "end": "00:09.3",
      "line": "His name is Tejas Chopra and the tool is called headroom.",
      "start": "00:05.8",
      "delivery": "Clear, informative introduction",
      "uncertain": false
    },
    {
      "end": "00:15.9",
      "line": "It sits between your app and the model. So every tool output, log and chat history passes through it and gets compressed before it reaches the model.",
      "start": "00:09.3",
      "delivery": "Structured technical explanation",
      "uncertain": false
    },
    {
      "end": "00:18.9",
      "line": "No summarizing and loss, just the same answer with way fewer tokens.",
      "start": "00:15.9",
      "delivery": "Reassuring, clarifying value proposition",
      "uncertain": false
    },
    {
      "end": "00:26.7",
      "line": "Setup is simple. 1 pip install, then run headroom-wrap-claude or start it as a proxy with zero code changes and you instantly see the savings.",
      "start": "00:18.9",
      "delivery": "Fast, actionable how-to walkthrough",
      "uncertain": false
    },
    {
      "end": "00:31.9",
      "line": "It works with Claude Code, Cursor and Copilot and crossed almost 30,000 stars on GitHub.",
      "start": "00:26.7",
      "delivery": "Strong closing social proof",
      "uncertain": false
    }
  ],
  "editing": {
    "cut_count": 15,
    "transitions": "Hard cuts timed precisely to word breaks",
    "caption_style": "Minimal dynamic centered subtitles synchronized to spoken words",
    "end_card_or_cta": "None; ends on GitHub star metric milestone",
    "average_shot_seconds": 2.1,
    "effects_and_overlays": "Split screen layout with rounded graphic cards over branded green textured background"
  },
  "teardown": {
    "hook": "This Indian engineer working at Netflix just built an open source tool that cuts your AI token usage by 60 to 95%.",
    "format": "talking-head",
    "mechanism": "specific_number",
    "structure": [
      "Hook introducing the engineer and impressive cost reduction claim",
      "Overview of the tool, how it sits as a compression proxy",
      "Fast setup demonstration via terminal command",
      "Compatibility list and social proof showing GitHub star growth"
    ],
    "hook_source": "transcript",
    "why_it_worked": "Credibility from a big tech engineer paired with a specific high-percentage cost savings hook drives immediate developer interest."
  },
  "on_screen_text": [
    {
      "end": "00:01.0",
      "kind": "hook",
      "text": "this indian engineer",
      "start": "00:00.0",
      "style": "Bold italic white font with dark outline",
      "source": "creator",
      "position": "middle_center"
    },
    {
      "end": "00:31.9",
      "kind": "label",
      "text": "30,000 stars",
      "start": "00:28.8",
      "style": "Black bold text next to gold star emoji",
      "source": "creator",
      "position": "top_center"
    }
  ],
  "structure_beats": [
    {
      "end": "00:05.8",
      "beat": "Credibility & Metric Hook",
      "start": "00:00.0",
      "purpose": "Capture AI developers by promising massive token cost reductions backed by a Netflix engineer"
    },
    {
      "end": "00:18.9",
      "beat": "How It Works",
      "start": "00:05.8",
      "purpose": "Explain mechanism clearly using visual diagrams without technical overload"
    },
    {
      "end": "00:26.7",
      "beat": "Implementation Demo",
      "start": "00:18.9",
      "purpose": "Demonstrate trivial single-command setup to remove developer friction"
    },
    {
      "end": "00:31.9",
      "beat": "Social Proof & Compatibility",
      "start": "00:26.7",
      "purpose": "Reinforce mainstream adoption with 30k stars and compatibility with Cursor and Copilot"
    }
  ],
  "recreation_blueprint": {
    "shot_template": [
      {
        "slot": "Hook",
        "visual": "Split screen with founder photo top and speaker bottom, cutting to company logo and savings chart",
        "seconds": 5.8,
        "voiceover": "This [Role] working at [Notable Company] just built an open source tool that cuts [Expense/Metric] by [Percentage]%.",
        "text_overlay": "Hook subtitle overlay"
      },
      {
        "slot": "Core Mechanism",
        "visual": "Diagrammatic animation showing data routing and lossless compression between input and destination",
        "seconds": 13.1,
        "voiceover": "His name is [Name] and the tool is [Product]. It sits between your app and [Service], compressing data without quality loss.",
        "text_overlay": "Tool name and diagram labels"
      },
      {
        "slot": "Quick Setup",
        "visual": "Terminal CLI run command followed by dashboard showing live savings",
        "seconds": 7.8,
        "voiceover": "Setup is simple. One install command, zero code changes, and you instantly see the savings.",
        "text_overlay": "Install command snippet"
      },
      {
        "slot": "Social Proof",
        "visual": "Tool integration checklist cutting to exponential GitHub star graph",
        "seconds": 5.2,
        "voiceover": "It works with [Tool A], [Tool B], and [Tool C], and crossed over [Number] stars on GitHub.",
        "text_overlay": "[Number] stars banner"
      }
    ],
    "what_can_change": [
      "Specific software product or metric targeted (e.g. database latency instead of AI tokens)",
      "Background colorway and aesthetic framing",
      "Presenter identity and delivery cadence"
    ],
    "template_summary": "High-velocity tech product teardown pairing an authority figure with a quantitative savings claim, followed by animated architecture diagrams and CLI install proof.",
    "what_must_be_kept": [
      "Immediate mention of authority company and extreme percentage improvement",
      "Visual diagram explaining the proxy architecture cleanly",
      "Terminal screen recording proving ease of setup"
    ],
    "generation_prompts": [
      {
        "prompt": "Split screen composition, top half outdoor portrait of smiling male software engineer, bottom half studio shot of bearded creator talking at microphone, minimalist office setting, clean lighting.",
        "shot_index": 1,
        "negative_prompt": "blurry, distorted face, saturated overspray, clutter"
      },
      {
        "prompt": "Clean modern UI dark mode dashboard showing token usage data table with animated bar graph filling from 60% to 95%, subtle glowing green accents, tech aesthetic.",
        "shot_index": 4,
        "negative_prompt": "noisy textures, pixelated, illegible text, low resolution"
      },
      {
        "prompt": "Minimal motion graphic diagram showing incoming message and log icons funneling into a central dark circle labeled with a logo and emerging compressed on white background.",
        "shot_index": 8,
        "negative_prompt": "messy illustration, chaotic particles, 3d clutter"
      },
      {
        "prompt": "Split screen, top displaying a clean GitHub Star History graph curving sharply upward toward 30,000 with star icon, bottom showing tech creator talking enthusiastically.",
        "shot_index": 15,
        "negative_prompt": "flat lighting, distorted graph line, compression artifacts"
      }
    ]
  },
  "props_apps_and_products": [
    {
      "name": "Headroom",
      "how_used": "Featured open-source software tool highlighted throughout video",
      "first_seen": "00:02.1"
    },
    {
      "name": "Claude Code",
      "how_used": "Terminal agent tool wrapped and compressed by Headroom",
      "first_seen": "00:21.0"
    },
    {
      "name": "GitHub",
      "how_used": "Platform showing repository code, profile, and star chart",
      "first_seen": "00:02.1"
    }
  ]
}