Outliers
@ai.with.andrew69k followers18d agotalking-head

This Bengaluru engineer built an app that detects potholes, finds the responsible contractor and files a complaint in 4 seconds.

Open on TikTok
6.2×its baseline
28k
views
4.9k
baseline
0.3×
engagement
1.0k
likes
178
saves
110
shares

Why it beat the baseline

hook · transcript

“This Bengaluru engineer built an app that detects potholes, finds the responsible contractor and files a complaint in 4 seconds.”

talking-headutility

  1. Hook introducing an engineer using AI to automate pothole complaints in 4 seconds
  2. Backstory on Gaurav Sen's commute problem and building the tool with OpenAI
  3. Technical breakdown of hardware and AI vision classification
  4. Civic data layer matching potholes with contractor warranty records
  5. Summary payoff showing automatic complaint generation per commute

Why it worked. Framed everyday commuter frustration into a high-satisfaction civic tech accountability story.

Grounded in the actual video and audio; analyzed by gemini-3.8-flash, cached.

Video analysis

Every shot, the spoken script, on-screen text, audio, editing and a recreation blueprint, from the actual video and audio.

talking-headshowcasehook: bold claim65s · 29 shots · 9:16speech · music · captionscta: none
Hook (spoken)This Bengaluru engineer built an app that detects potholes, finds the responsible contractor and files a complaint in 4 seconds.
Hook (on screen)ai just declared WAR ON POTHOLES
Why it stops the scrollPromises instant retribution against bad road infrastructure using AI.
TechniqueCombines a viral civic pain point with an impossible-sounding automated tech solution.
FormatSplit-screen talking head with top B-roll overlay
SettingMinimalist home office studio
SubjectMale tech commentator speaking to camera at a desk
LookBlack crewneck t-shirt, dark beard and styled dark hair
Lighting / gradeSoft key light from front-left with subtle rim lighting; Neutral, clean digital tech palette
Pacing / toneFast, rhythmic cuts matching voiceover cadence; Engaging, authoritative, narrative-driven
AudioClear, fast-paced male North American voiceover. Music: Million Dolla Hip Hop instrumental loop in background. SFX: Subtle swooshes on graphics cuts.
Editing28 cuts, 2.33s avg; Hard cuts throughout; captions: Single/two-word animated subtitle overlay in bottom center
Topicsai, civic tech, computer vision, automation, india
Products / appsQubo Dashcam (00:22.7: Hardware device collecting driving video and sensor data); ADXL335 Accelerometer (00:26.0: Sensor graphic explaining bump detection); OpenAI Codex / GPT Vision (00:16.8: AI models used to analyze footage and build the application); BBMP Pothole Report Portal (00:45.7: Public municipal data dashboard queried for road contract data)

Structure · 65s

  1. 00:00.0–00:06.6Core Value Hook. State the high-utility promise of instant 4-second contractor complaints.
  2. 00:06.6–00:22.7Commuter Origin Story. Establish personal frustration and motivation to build an AI accountability tool.
  3. 00:22.7–00:36.4Hardware & Vision Detection. Break down dashcam, accelerometer trigger, and AI vision size classification.
  4. 00:36.4–00:56.4Civic Accountability Hack. Reveal warranty legal loophole and database cross-referencing.
  5. 00:56.4–01:05.37Payoff Resolution. Deliver circular payoff showing 12 potholes transformed into 12 ready complaints.

Shot list

#timeshotwhat is visiblespokenon screen
100:00.0–00:03.0split-screen
Fixed eye-level front; Static		Top screen shows road footage with red detection boxes and Gaurav Sen; bottom shows presenter speaking
Screen: Pothole detection overlay with labels 'roadpothole'		This Bengaluru engineer built an app that detects potholes,ai just declared WAR ON POTHOLES
200:03.0–00:04.7medium
Eye-level news broadcast; Static		Nitin Gadkari seated behind laptop in front of Indian flag and ministry signfinds the responsible contractor
300:04.7–00:06.6close-up
Mounted eye-level; Static		Smartphone displaying app screen showing 'Complaint Ready 4 seconds'
Screen: Green checkmark, 'Complaint Ready 4 seconds', contractor details		and files a complaint in 4 seconds.Complaint Ready 4 seconds
400:06.6–00:07.7close-up
Car interior dash mount; Handheld vibration		Gaurav Sen wearing glasses driving at nightHere's how it works. Gaurav Sen was
500:07.7–00:09.9split-screen
Fixed eye-level; Static		Highway driving footage on top, presenter speaking below
Screen: Highway dashcam		driving to work in Bengaluru and hitting
600:09.9–00:11.0split-screen
Fixed eye-level; Static		Cracked pothole road top, presenter below
Screen: B-roll of damaged rural road		12 potholes every day.
700:11.0–00:12.7medium
Street level tracking; Pan right		Car wheel splashing water in road craterMost people would just suffer.
800:12.7–00:15.8medium
Eye-level tripod; Static		Full view of creator at deskbut he decided to hold people responsible. So he partnered with OpenAI
900:15.8–00:16.8medium
Side angle eye-level; Static		Gaurav Sen typing on laptopand built an app
1000:16.8–00:18.1split-screen
Eye-level; Static		Word 'Codex' in blue gradient upper screen, presenter below
Screen: Text 'Codex'		with Codex that not onlyCodex
1100:18.1–00:20.9medium
Street-level side angle; Handheld pan		SUV tire sinking into deep roadside pothole next to scooterdetects potholes, but also finds who is responsible
1200:20.9–00:22.7medium
Handheld eye-level; Subtle handheld drift		Construction machinery and obscured worker portrait
Screen: Large red '?' over face		and files the complaint. Here's how.?
1300:22.7–00:26.0close-up
Windshield mount macro; Graphic exploded overlay		Dashcam illuminated in red wireframe with floating sensor graphics
Screen: Text 'KIND OF DASHCAM' and hardware callout		He put a dashcam on his car with GPSKIND OF DASHCAM
1400:26.0–00:26.9split-screen
Eye-level; Static		Accelerometer circuit board top, presenter below
Screen: ADXL335 chip on grey backdrop		and an accelerometer.Accelerometer · ADXL335
1500:26.9–00:28.1close-up
Ground-level high speed; Static		Mud spray explosion under car wheel impactIt knows when the car
1600:28.1–00:29.0POV
Dashboard windshield mount; Car moving forward		Wipers and uneven rural road surfacehits a bump.
1700:29.0–00:31.5split-screen
Eye-level; Static		Pothole computer vision bounding box top, presenter gesturing wide below
Screen: CV bounding box 'pothole 0.55'		The system records the road and detects potholes. The footagepothole 0.55
1800:31.5–00:33.8split-screen
Eye-level; Static		Node.js/TypeScript code for OpenAI vision model on top, presenter below
Screen: Code snippet importing 'gpt-5' vision models		is processed using AI vision modelsVision Models · const VISION_MODEL = "gpt-5";
1900:33.8–00:35.0medium
Street level pan; Pan left		Minivan driving past damaged road intersectionto classify potholes as
2000:35.0–00:36.4graphic
Graphic UI screen; Static		Potholes highlighted in red spray paint with labels
Screen: Images marked 'Small' and 'Medium' with red spray highlights		small, medium, or large.Small · Medium
2100:36.4–00:37.8medium
Eye-level tripod; Static		Presenter smiling slightly, opening the next hook sectionNow here's where it gets clever.
2200:37.8–00:40.4split-screen
Eye-level; Static		Heavy construction truck splashing road water top, presenter speaking below
Screen: B-roll of heavy truck on broken road		Many Indian roads are still under construction warranty.
2300:40.4–00:42.5split-screen
Eye-level; Static		Commuters on motorbikes avoiding potholes top, presenter below
Screen: B-roll of Indian traffic on bad road		So the contractor who built the road
2400:42.5–00:45.7split-screen
Eye-level; Static		Flooded street surface top, presenter with both hands raised below
Screen: Pothole puddle B-roll		is actually responsible for fixing the potholes.
2500:45.7–00:49.7close-up
Graphic UI mobile frame; Scroll down animation		App interface showing 36,854 identified potholes and zone pie chart
Screen: BBMP dashboard showing metrics: '36854 IDENTIFIED POTHOLES', zone chart		Gaurav's app searches through 2,900 government contractsPotholes detailed Daily Report · 36854 IDENTIFIED POTHOLES · BBMP POTHOLES
2600:49.7–00:56.4split-screen
Eye-level; Static		Mobile report scrolling through zones (Bommanahalli, Dasarahalli) and bar chart top, presenter below
Screen: Zone wise identified potholes bar graph and data table		and finds the photograph, exact location, tender number, and the officer responsible for the road, all within 4 seconds.ZONE WISE REPORT · Zone wise Identified Potholes
2700:56.4–00:58.8medium
Eye-level tripod; Static		Presenter speaking with open hand gesturesThen the app makes a structured complaint-ready record.
2800:58.8–01:02.5POV
Dashboard windshield mount; Car motion		Cracked two-lane country road with multiple patchworksSo on one drive to work, if you hit 12 potholes,
2901:02.5–01:05.37split-screen
Eye-level; Static		Phone mount in car showing filled contractor form top, presenter speaking below
Screen: Form showing 'Complaint Ready 4 seconds' with tender and contractor ID		you have 12 complaints ready by the end of your ride.Complaint Ready 4 seconds

Script

  1. 00:00.0–00:03.0“This Bengaluru engineer built an app that detects potholes,” · High energy hook delivery
  2. 00:03.0–00:04.7“finds the responsible contractor” · Emphasized pacing
  3. 00:04.7–00:06.6“and files a complaint in 4 seconds.” · Punchy punchline cadence
  4. 00:06.6–00:07.7“Here's how it works.” · Quick transition bridge
  5. 00:07.7–00:11.0“Gaurav Sen was driving to work in Bengaluru and hitting 12 potholes every day.” · Storytelling setup
  6. 00:11.0–00:12.7“Most people would just suffer.” · Contrasting statement
  7. 00:12.7–00:14.7“but he decided to hold people responsible.” · Determined tone
  8. 00:14.7–00:18.1“So he partnered with OpenAI and built an app with Codex that not only” · Technical backstory setup
  9. 00:18.1–00:20.9“detects potholes, but also finds who is responsible” · Rhythmic repetition of features
  10. 00:20.9–00:22.7“and files the complaint. Here's how.” · Brisk narrative reset
  11. 00:22.7–00:26.9“He put a dashcam on his car with GPS and an accelerometer.” · Instructive technical breakdown
  12. 00:26.9–00:29.0“It knows when the car hits a bump.” · Direct and plain explanation
  13. 00:29.0–00:31.5“The system records the road and detects potholes.” · Clear operational step
  14. 00:31.5–00:36.4“The footage is processed using AI vision models to classify potholes as small, medium or large.” · Informative tech delivery
  15. 00:36.4–00:37.8“Now here's where it gets clever.” · Intrigue-building pivot
  16. 00:37.8–00:40.4“Many Indian roads are still under construction warranty.” · Educational insight
  17. 00:40.4–00:45.7“So the contractor who built the road is actually responsible for fixing the potholes.” · Emphasized revelation
  18. 00:45.7–00:49.7“Gaurav's app searches through 2,900 government contracts” · Specific numerical proof
  19. 00:49.7–00:56.4“and finds the photograph, exact location, tender number and the officer responsible for the road, all within 4 seconds.” · Rapid feature listing with payoff timing
  20. 00:56.4–00:58.8“Then the app makes a structured complaint-ready record.” · Concluding system summary
  21. 00:58.8–01:05.37“So on one drive to work, if you hit 12 potholes, you have 12 complaints ready by the end of your ride.” · Satisfying circular resolution

On-screen text

  • 00:00.0–00:03.0“ai just declared WAR ON POTHOLES” · hook · center top · Bold white sans-serif uppercase centered overlay
  • 00:04.7–00:06.6“Complaint Ready 4 seconds” · embedded · center UI · Green bold heading with round checkmark icon
  • 00:16.8–00:18.1“Codex” · label · center top · Large white sans-serif on blue gradient
  • 00:26.0–00:26.9“Accelerometer ADXL335” · label · center top · Black bold title above circuit board component
  • 00:29.0–00:31.5“pothole 0.55” · embedded · center road · Red bounding box label over detected hole
  • 00:31.5–00:33.8“Vision Models” · label · top center · White sans-serif title over dark code screen
  • 00:35.0–00:36.4“Small / Medium” · label · center UI · Clean grey category titles over pothole photos
  • 00:45.7–00:56.4“Potholes detailed Daily Report” · embedded · top of mobile screen · White text on black app header bar

Recreation blueprint

Civic innovation breakdown using split-screen commentary, sensor explanation, and database payoff.

slotsvisualvoiceoveroverlay
Hook6.6Split screen: top demo of problem detection; bottom creator talking head.This [Location] [Profession] built an app that solves [Common Pain] in [Short Time].[TOPIC] JUST DECLARED WAR ON [PROBLEM]
Backstory16.1B-roll of daily commuter struggle and coding on laptop.[Founder] suffered from [Pain] daily, so they used [AI Platform] to build a solution.[AI Tool Name]
Tech How-It-Works13.7Product hardware close-up and code snippet classification.Here is how it works: sensor triggers capture, processed by AI vision models.[Sensor / Model Label]
System Secret20Dashboard UI cross-referencing public legal or government registry records.Here is the clever part: matching faults to public records and responsible parties automatically.[Database / Metric Name]
Payoff Outro8.97Dashcam driving footage cut to completed report on smartphone.So on one [Action], you have [Number] automated solutions ready by the time you arrive.[Completed Action Status]

Keep

  • Split-screen layout pairing talking head with contextual B-roll
  • Specific metrics (time in seconds, contract counts, daily bumps)
  • Technical credibility through real code and hardware names

Can change

  • Target municipal or consumer problem (e.g., parking fines, delayed trains, illegal dumping)
  • Specific hardware sensors and AI model providers used
Generation prompts per shot (3)
  1. Shot 1. Split screen 9:16 vertical video. Top half shows front car dashcam road footage with red bounding boxes tracking road craters. Bottom half shows young male tech creator talking enthusiastically to camera in a modern desk studio.
    Negative: blurry, low resolution, saturated cartoon colors, cluttered background
  2. Shot 13. Cinematic close-up of compact automotive dashcam mounted on car windshield with futuristic red holographic outline highlighting internal sensors and GPS antenna.
    Negative: cracked glass, messy wiring, dark underexposed lighting
  3. Shot 25. Clean mobile UI mockup of civic audit dashboard displaying pie charts, zone metrics, and government contract lists on clean modern dark/light mode interface.
    Negative: unreadable text, distorted layout, noisy artifacts
Raw JSON (for a video pipeline)
{
  "hook": {
    "spoken": "This Bengaluru engineer built an app that detects potholes, finds the responsible contractor and files a complaint in 4 seconds.",
    "ends_at": "00:06.6",
    "on_screen": "ai just declared WAR ON POTHOLES",
    "starts_at": "00:00.0",
    "technique": "bold_claim",
    "technique_note": "Combines a viral civic pain point with an impossible-sounding automated tech solution.",
    "why_it_stops_the_scroll": "Promises instant retribution against bad road infrastructure using AI."
  },
  "tags": {
    "topics": [
      "ai",
      "civic tech",
      "computer vision",
      "automation",
      "india"
    ],
    "cta_type": "none",
    "language": "en",
    "has_music": true,
    "has_speech": true,
    "content_type": "showcase",
    "products_or_apps": [
      "OpenAI Codex",
      "Qubo Dashcam"
    ],
    "has_burned_captions": true
  },
  "audio": {
    "music": "Million Dolla Hip Hop instrumental loop in background",
    "voice": "Clear, fast-paced male North American voiceover",
    "ambient": "None audible under mix",
    "mix_notes": "Voice track sits prominently on top of low-volume hip hop beat",
    "sound_effects": "Subtle swooshes on graphics cuts"
  },
  "shots": [
    {
      "end": "00:03.0",
      "index": 1,
      "start": "00:00.0",
      "camera": "Fixed eye-level front",
      "framing": "Top: computer vision pothole detection and engineer portrait; Bottom: medium talking head",
      "movement": "Static",
      "shot_type": "split-screen",
      "transition_in": "None",
      "screen_content": "Pothole detection overlay with labels 'roadpothole'",
      "subject_action": "Presenter speaks to camera gesturing with hands",
      "what_is_visible": "Top screen shows road footage with red detection boxes and Gaurav Sen; bottom shows presenter speaking",
      "duration_seconds": 3,
      "spoken_during_shot": "This Bengaluru engineer built an app that detects potholes,",
      "on_screen_text_during_shot": [
        "ai just declared WAR ON POTHOLES"
      ],
      "facial_expression_or_gesture": "Engaged, open-handed gestures"
    },
    {
      "end": "00:04.7",
      "index": 2,
      "start": "00:03.0",
      "camera": "Eye-level news broadcast",
      "framing": "Medium shot of Indian government minister at conference table",
      "movement": "Static",
      "shot_type": "medium",
      "transition_in": "Cut",
      "screen_content": "None",
      "subject_action": "Speaking into conference microphone",
      "what_is_visible": "Nitin Gadkari seated behind laptop in front of Indian flag and ministry sign",
      "duration_seconds": 1.7,
      "spoken_during_shot": "finds the responsible contractor",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "Serious official expression"
    },
    {
      "end": "00:06.6",
      "index": 3,
      "start": "00:04.7",
      "camera": "Mounted eye-level",
      "framing": "Close-up of smartphone in dashboard mount",
      "movement": "Static",
      "shot_type": "close-up",
      "transition_in": "Cut",
      "screen_content": "Green checkmark, 'Complaint Ready 4 seconds', contractor details",
      "subject_action": "App interface populates with tender data",
      "what_is_visible": "Smartphone displaying app screen showing 'Complaint Ready 4 seconds'",
      "duration_seconds": 1.9,
      "spoken_during_shot": "and files a complaint in 4 seconds.",
      "on_screen_text_during_shot": [
        "Complaint Ready 4 seconds"
      ],
      "facial_expression_or_gesture": "N/A"
    },
    {
      "end": "00:07.7",
      "index": 4,
      "start": "00:06.6",
      "camera": "Car interior dash mount",
      "framing": "Close-up of engineer driving at night",
      "movement": "Handheld vibration",
      "shot_type": "close-up",
      "transition_in": "Cut",
      "screen_content": "None",
      "subject_action": "Looking ahead while driving",
      "what_is_visible": "Gaurav Sen wearing glasses driving at night",
      "duration_seconds": 1.1,
      "spoken_during_shot": "Here's how it works. Gaurav Sen was",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "Focused"
    },
    {
      "end": "00:09.9",
      "index": 5,
      "start": "00:07.7",
      "camera": "Fixed eye-level",
      "framing": "Top: dashcam sunrise road footage; Bottom: medium talking head",
      "movement": "Static",
      "shot_type": "split-screen",
      "transition_in": "Cut",
      "screen_content": "Highway dashcam",
      "subject_action": "Presenter explains context",
      "what_is_visible": "Highway driving footage on top, presenter speaking below",
      "duration_seconds": 2.2,
      "spoken_during_shot": "driving to work in Bengaluru and hitting",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "Explaining with open palms"
    },
    {
      "end": "00:11.0",
      "index": 6,
      "start": "00:09.9",
      "camera": "Fixed eye-level",
      "framing": "Top: rural pothole road POV; Bottom: medium talking head",
      "movement": "Static",
      "shot_type": "split-screen",
      "transition_in": "Cut",
      "screen_content": "B-roll of damaged rural road",
      "subject_action": "Presenter raises hands to emphasize count",
      "what_is_visible": "Cracked pothole road top, presenter below",
      "duration_seconds": 1.1,
      "spoken_during_shot": "12 potholes every day.",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "Surprised expression"
    },
    {
      "end": "00:12.7",
      "index": 7,
      "start": "00:11.0",
      "camera": "Street level tracking",
      "framing": "Monochrome street view of white SUV hitting pothole puddle",
      "movement": "Pan right",
      "shot_type": "medium",
      "transition_in": "Cut",
      "screen_content": "None",
      "subject_action": "Vehicle drives through damaged asphalt puddle",
      "what_is_visible": "Car wheel splashing water in road crater",
      "duration_seconds": 1.7,
      "spoken_during_shot": "Most people would just suffer.",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "N/A"
    },
    {
      "end": "00:15.8",
      "index": 8,
      "start": "00:12.7",
      "camera": "Eye-level tripod",
      "framing": "Presenter at desk with microphone and laptop",
      "movement": "Static",
      "shot_type": "medium",
      "transition_in": "Cut",
      "screen_content": "None",
      "subject_action": "Speaks and gestures firmly",
      "what_is_visible": "Full view of creator at desk",
      "duration_seconds": 3.1,
      "spoken_during_shot": "but he decided to hold people responsible. So he partnered with OpenAI",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "Direct, resolute"
    },
    {
      "end": "00:16.8",
      "index": 9,
      "start": "00:15.8",
      "camera": "Side angle eye-level",
      "framing": "Medium shot of engineer working on laptop at modern desk",
      "movement": "Static",
      "shot_type": "medium",
      "transition_in": "Cut",
      "screen_content": "None",
      "subject_action": "Coding at desk",
      "what_is_visible": "Gaurav Sen typing on laptop",
      "duration_seconds": 1,
      "spoken_during_shot": "and built an app",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "Concentrated"
    },
    {
      "end": "00:18.1",
      "index": 10,
      "start": "00:16.8",
      "camera": "Eye-level",
      "framing": "Top: gradient background with text 'Codex'; Bottom: presenter",
      "movement": "Static",
      "shot_type": "split-screen",
      "transition_in": "Cut",
      "screen_content": "Text 'Codex'",
      "subject_action": "Explains tech stack",
      "what_is_visible": "Word 'Codex' in blue gradient upper screen, presenter below",
      "duration_seconds": 1.3,
      "spoken_during_shot": "with Codex that not only",
      "on_screen_text_during_shot": [
        "Codex"
      ],
      "facial_expression_or_gesture": "Animated speaking"
    },
    {
      "end": "00:20.9",
      "index": 11,
      "start": "00:18.1",
      "camera": "Street-level side angle",
      "framing": "White SUV driving into massive road depression",
      "movement": "Handheld pan",
      "shot_type": "medium",
      "transition_in": "Cut",
      "screen_content": "None",
      "subject_action": "Car dips heavily into hole",
      "what_is_visible": "SUV tire sinking into deep roadside pothole next to scooter",
      "duration_seconds": 2.8,
      "spoken_during_shot": "detects potholes, but also finds who is responsible",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "N/A"
    },
    {
      "end": "00:22.7",
      "index": 12,
      "start": "00:20.9",
      "camera": "Handheld eye-level",
      "framing": "Contractor standing by yellow road roller with red question mark over face",
      "movement": "Subtle handheld drift",
      "shot_type": "medium",
      "transition_in": "Cut",
      "screen_content": "Large red '?' over face",
      "subject_action": "Posing next to steamroller",
      "what_is_visible": "Construction machinery and obscured worker portrait",
      "duration_seconds": 1.8,
      "spoken_during_shot": "and files the complaint. Here's how.",
      "on_screen_text_during_shot": [
        "?"
      ],
      "facial_expression_or_gesture": "Face masked by question mark graphic"
    },
    {
      "end": "00:26.0",
      "index": 13,
      "start": "00:22.7",
      "camera": "Windshield mount macro",
      "framing": "Close-up of Qubo dashcam on windshield with 3D animation",
      "movement": "Graphic exploded overlay",
      "shot_type": "close-up",
      "transition_in": "Cut",
      "screen_content": "Text 'KIND OF DASHCAM' and hardware callout",
      "subject_action": "Visual breakdown of dashcam hardware",
      "what_is_visible": "Dashcam illuminated in red wireframe with floating sensor graphics",
      "duration_seconds": 3.3,
      "spoken_during_shot": "He put a dashcam on his car with GPS",
      "on_screen_text_during_shot": [
        "KIND OF DASHCAM"
      ],
      "facial_expression_or_gesture": "N/A"
    },
    {
      "end": "00:26.9",
      "index": 14,
      "start": "00:26.0",
      "camera": "Eye-level",
      "framing": "Top: ADXL335 accelerometer chip graphic; Bottom: presenter",
      "movement": "Static",
      "shot_type": "split-screen",
      "transition_in": "Cut",
      "screen_content": "ADXL335 chip on grey backdrop",
      "subject_action": "Presenter explains sensor trigger",
      "what_is_visible": "Accelerometer circuit board top, presenter below",
      "duration_seconds": 0.9,
      "spoken_during_shot": "and an accelerometer.",
      "on_screen_text_during_shot": [
        "Accelerometer",
        "ADXL335"
      ],
      "facial_expression_or_gesture": "Nodding"
    },
    {
      "end": "00:28.1",
      "index": 15,
      "start": "00:26.9",
      "camera": "Ground-level high speed",
      "framing": "Car tire splashing muddy water from deep pothole",
      "movement": "Static",
      "shot_type": "close-up",
      "transition_in": "Cut",
      "screen_content": "None",
      "subject_action": "Car strikes bump",
      "what_is_visible": "Mud spray explosion under car wheel impact",
      "duration_seconds": 1.2,
      "spoken_during_shot": "It knows when the car",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "N/A"
    },
    {
      "end": "00:29.0",
      "index": 16,
      "start": "00:28.1",
      "camera": "Dashboard windshield mount",
      "framing": "Rainy road dashcam view hitting bumps",
      "movement": "Car moving forward",
      "shot_type": "POV",
      "transition_in": "Cut",
      "screen_content": "None",
      "subject_action": "Vehicle navigating rough terrain",
      "what_is_visible": "Wipers and uneven rural road surface",
      "duration_seconds": 0.9,
      "spoken_during_shot": "hits a bump.",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "N/A"
    },
    {
      "end": "00:31.5",
      "index": 17,
      "start": "00:29.0",
      "camera": "Eye-level",
      "framing": "Top: road POV with red bounding box labeled 'pothole 0.55'; Bottom: presenter",
      "movement": "Static",
      "shot_type": "split-screen",
      "transition_in": "Cut",
      "screen_content": "CV bounding box 'pothole 0.55'",
      "subject_action": "Presenter explains automated detection",
      "what_is_visible": "Pothole computer vision bounding box top, presenter gesturing wide below",
      "duration_seconds": 2.5,
      "spoken_during_shot": "The system records the road and detects potholes. The footage",
      "on_screen_text_during_shot": [
        "pothole 0.55"
      ],
      "facial_expression_or_gesture": "Expressive hand movements"
    },
    {
      "end": "00:33.8",
      "index": 18,
      "start": "00:31.5",
      "camera": "Eye-level",
      "framing": "Top: code editor showing GPT-5 vision API calls; Bottom: presenter",
      "movement": "Static",
      "shot_type": "split-screen",
      "transition_in": "Cut",
      "screen_content": "Code snippet importing 'gpt-5' vision models",
      "subject_action": "Presenter explains image processing",
      "what_is_visible": "Node.js/TypeScript code for OpenAI vision model on top, presenter below",
      "duration_seconds": 2.3,
      "spoken_during_shot": "is processed using AI vision models",
      "on_screen_text_during_shot": [
        "Vision Models",
        "const VISION_MODEL = \"gpt-5\";"
      ],
      "facial_expression_or_gesture": "Active explanation"
    },
    {
      "end": "00:35.0",
      "index": 19,
      "start": "00:33.8",
      "camera": "Street level pan",
      "framing": "City traffic navigating around street potholes",
      "movement": "Pan left",
      "shot_type": "medium",
      "transition_in": "Cut",
      "screen_content": "None",
      "subject_action": "Traffic maneuvers around road faults",
      "what_is_visible": "Minivan driving past damaged road intersection",
      "duration_seconds": 1.2,
      "spoken_during_shot": "to classify potholes as",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "N/A"
    },
    {
      "end": "00:36.4",
      "index": 20,
      "start": "00:35.0",
      "camera": "Graphic UI screen",
      "framing": "Vertical UI displaying 'Small' and 'Medium' marked potholes",
      "movement": "Static",
      "shot_type": "graphic",
      "transition_in": "Cut",
      "screen_content": "Images marked 'Small' and 'Medium' with red spray highlights",
      "subject_action": "Graphic demonstration",
      "what_is_visible": "Potholes highlighted in red spray paint with labels",
      "duration_seconds": 1.4,
      "spoken_during_shot": "small, medium, or large.",
      "on_screen_text_during_shot": [
        "Small",
        "Medium"
      ],
      "facial_expression_or_gesture": "N/A"
    },
    {
      "end": "00:37.8",
      "index": 21,
      "start": "00:36.4",
      "camera": "Eye-level tripod",
      "framing": "Presenter at desk",
      "movement": "Static",
      "shot_type": "medium",
      "transition_in": "Cut",
      "screen_content": "None",
      "subject_action": "Leans forward engaging viewer",
      "what_is_visible": "Presenter smiling slightly, opening the next hook section",
      "duration_seconds": 1.4,
      "spoken_during_shot": "Now here's where it gets clever.",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "Intrigued, smiling"
    },
    {
      "end": "00:40.4",
      "index": 22,
      "start": "00:37.8",
      "camera": "Eye-level",
      "framing": "Top: yellow dump truck driving through waterlogged pothole; Bottom: presenter",
      "movement": "Static",
      "shot_type": "split-screen",
      "transition_in": "Cut",
      "screen_content": "B-roll of heavy truck on broken road",
      "subject_action": "Presenter explains warranty regulations",
      "what_is_visible": "Heavy construction truck splashing road water top, presenter speaking below",
      "duration_seconds": 2.6,
      "spoken_during_shot": "Many Indian roads are still under construction warranty.",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "Explaining with hand motions"
    },
    {
      "end": "00:42.5",
      "index": 23,
      "start": "00:40.4",
      "camera": "Eye-level",
      "framing": "Top: crowded scooters maneuvering through water-filled pothole; Bottom: presenter",
      "movement": "Static",
      "shot_type": "split-screen",
      "transition_in": "Cut",
      "screen_content": "B-roll of Indian traffic on bad road",
      "subject_action": "Presenter articulates legal obligation",
      "what_is_visible": "Commuters on motorbikes avoiding potholes top, presenter below",
      "duration_seconds": 2.1,
      "spoken_during_shot": "So the contractor who built the road",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "Emphatic delivery"
    },
    {
      "end": "00:45.7",
      "index": 24,
      "start": "00:42.5",
      "camera": "Eye-level",
      "framing": "Top: muddy flooded road surface; Bottom: presenter",
      "movement": "Static",
      "shot_type": "split-screen",
      "transition_in": "Cut",
      "screen_content": "Pothole puddle B-roll",
      "subject_action": "Presenter gestures outward",
      "what_is_visible": "Flooded street surface top, presenter with both hands raised below",
      "duration_seconds": 3.2,
      "spoken_during_shot": "is actually responsible for fixing the potholes.",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "Direct emphasis"
    },
    {
      "end": "00:49.7",
      "index": 25,
      "start": "00:45.7",
      "camera": "Graphic UI mobile frame",
      "framing": "Mobile phone frame showing BBMP Potholes Detailed Daily Report dashboard",
      "movement": "Scroll down animation",
      "shot_type": "close-up",
      "transition_in": "Cut",
      "screen_content": "BBMP dashboard showing metrics: '36854 IDENTIFIED POTHOLES', zone chart",
      "subject_action": "App UI scrolls to reveal data charts",
      "what_is_visible": "App interface showing 36,854 identified potholes and zone pie chart",
      "duration_seconds": 4,
      "spoken_during_shot": "Gaurav's app searches through 2,900 government contracts",
      "on_screen_text_during_shot": [
        "Potholes detailed Daily Report",
        "36854 IDENTIFIED POTHOLES",
        "BBMP POTHOLES"
      ],
      "facial_expression_or_gesture": "N/A"
    },
    {
      "end": "00:56.4",
      "index": 26,
      "start": "00:49.7",
      "camera": "Eye-level",
      "framing": "Top: scrolling mobile report showing zone-wise potholes and contract stats; Bottom: presenter",
      "movement": "Static",
      "shot_type": "split-screen",
      "transition_in": "Cut",
      "screen_content": "Zone wise identified potholes bar graph and data table",
      "subject_action": "Presenter lists data elements retrieved",
      "what_is_visible": "Mobile report scrolling through zones (Bommanahalli, Dasarahalli) and bar chart top, presenter below",
      "duration_seconds": 6.7,
      "spoken_during_shot": "and finds the photograph, exact location, tender number, and the officer responsible for the road, all within 4 seconds.",
      "on_screen_text_during_shot": [
        "ZONE WISE REPORT",
        "Zone wise Identified Potholes"
      ],
      "facial_expression_or_gesture": "Counting points on hands"
    },
    {
      "end": "00:58.8",
      "index": 27,
      "start": "00:56.4",
      "camera": "Eye-level tripod",
      "framing": "Presenter at desk",
      "movement": "Static",
      "shot_type": "medium",
      "transition_in": "Cut",
      "screen_content": "None",
      "subject_action": "Concludes the system summary",
      "what_is_visible": "Presenter speaking with open hand gestures",
      "duration_seconds": 2.4,
      "spoken_during_shot": "Then the app makes a structured complaint-ready record.",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "Clear and confident"
    },
    {
      "end": "01:02.5",
      "index": 28,
      "start": "00:58.8",
      "camera": "Dashboard windshield mount",
      "framing": "Wiper wiping windshield while driving on bumpy road with severe asphalt patches",
      "movement": "Car motion",
      "shot_type": "POV",
      "transition_in": "Cut",
      "screen_content": "None",
      "subject_action": "Driving along damaged road",
      "what_is_visible": "Cracked two-lane country road with multiple patchworks",
      "duration_seconds": 3.7,
      "spoken_during_shot": "So on one drive to work, if you hit 12 potholes,",
      "on_screen_text_during_shot": [],
      "facial_expression_or_gesture": "N/A"
    },
    {
      "end": "01:05.37",
      "index": 29,
      "start": "01:02.5",
      "camera": "Eye-level",
      "framing": "Top: phone mount with complaint ready screen; Bottom: presenter",
      "movement": "Static",
      "shot_type": "split-screen",
      "transition_in": "Cut",
      "screen_content": "Form showing 'Complaint Ready 4 seconds' with tender and contractor ID",
      "subject_action": "Presenter summarizes concluding line",
      "what_is_visible": "Phone mount in car showing filled contractor form top, presenter speaking below",
      "duration_seconds": 2.87,
      "spoken_during_shot": "you have 12 complaints ready by the end of your ride.",
      "on_screen_text_during_shot": [
        "Complaint Ready 4 seconds"
      ],
      "facial_expression_or_gesture": "Satisfied smile and open hands"
    }
  ],
  "video": {
    "pacing": "Fast, rhythmic cuts matching voiceover cadence",
    "setting": "Minimalist home office studio",
    "subject": "Male tech commentator speaking to camera at a desk",
    "lighting": "Soft key light from front-left with subtle rim lighting",
    "color_grade": "Neutral, clean digital tech palette",
    "orientation": "vertical",
    "aspect_ratio": "9:16",
    "overall_format": "Split-screen talking head with top B-roll overlay",
    "tone_and_energy": "Engaging, authoritative, narrative-driven",
    "duration_seconds": 65.37,
    "wardrobe_and_look": "Black crewneck t-shirt, dark beard and styled dark hair"
  },
  "script": [
    {
      "end": "00:03.0",
      "line": "This Bengaluru engineer built an app that detects potholes,",
      "start": "00:00.0",
      "delivery": "High energy hook delivery",
      "uncertain": false
    },
    {
      "end": "00:04.7",
      "line": "finds the responsible contractor",
      "start": "00:03.0",
      "delivery": "Emphasized pacing",
      "uncertain": false
    },
    {
      "end": "00:06.6",
      "line": "and files a complaint in 4 seconds.",
      "start": "00:04.7",
      "delivery": "Punchy punchline cadence",
      "uncertain": false
    },
    {
      "end": "00:07.7",
      "line": "Here's how it works.",
      "start": "00:06.6",
      "delivery": "Quick transition bridge",
      "uncertain": false
    },
    {
      "end": "00:11.0",
      "line": "Gaurav Sen was driving to work in Bengaluru and hitting 12 potholes every day.",
      "start": "00:07.7",
      "delivery": "Storytelling setup",
      "uncertain": false
    },
    {
      "end": "00:12.7",
      "line": "Most people would just suffer.",
      "start": "00:11.0",
      "delivery": "Contrasting statement",
      "uncertain": false
    },
    {
      "end": "00:14.7",
      "line": "but he decided to hold people responsible.",
      "start": "00:12.7",
      "delivery": "Determined tone",
      "uncertain": false
    },
    {
      "end": "00:18.1",
      "line": "So he partnered with OpenAI and built an app with Codex that not only",
      "start": "00:14.7",
      "delivery": "Technical backstory setup",
      "uncertain": false
    },
    {
      "end": "00:20.9",
      "line": "detects potholes, but also finds who is responsible",
      "start": "00:18.1",
      "delivery": "Rhythmic repetition of features",
      "uncertain": false
    },
    {
      "end": "00:22.7",
      "line": "and files the complaint. Here's how.",
      "start": "00:20.9",
      "delivery": "Brisk narrative reset",
      "uncertain": false
    },
    {
      "end": "00:26.9",
      "line": "He put a dashcam on his car with GPS and an accelerometer.",
      "start": "00:22.7",
      "delivery": "Instructive technical breakdown",
      "uncertain": false
    },
    {
      "end": "00:29.0",
      "line": "It knows when the car hits a bump.",
      "start": "00:26.9",
      "delivery": "Direct and plain explanation",
      "uncertain": false
    },
    {
      "end": "00:31.5",
      "line": "The system records the road and detects potholes.",
      "start": "00:29.0",
      "delivery": "Clear operational step",
      "uncertain": false
    },
    {
      "end": "00:36.4",
      "line": "The footage is processed using AI vision models to classify potholes as small, medium or large.",
      "start": "00:31.5",
      "delivery": "Informative tech delivery",
      "uncertain": false
    },
    {
      "end": "00:37.8",
      "line": "Now here's where it gets clever.",
      "start": "00:36.4",
      "delivery": "Intrigue-building pivot",
      "uncertain": false
    },
    {
      "end": "00:40.4",
      "line": "Many Indian roads are still under construction warranty.",
      "start": "00:37.8",
      "delivery": "Educational insight",
      "uncertain": false
    },
    {
      "end": "00:45.7",
      "line": "So the contractor who built the road is actually responsible for fixing the potholes.",
      "start": "00:40.4",
      "delivery": "Emphasized revelation",
      "uncertain": false
    },
    {
      "end": "00:49.7",
      "line": "Gaurav's app searches through 2,900 government contracts",
      "start": "00:45.7",
      "delivery": "Specific numerical proof",
      "uncertain": false
    },
    {
      "end": "00:56.4",
      "line": "and finds the photograph, exact location, tender number and the officer responsible for the road, all within 4 seconds.",
      "start": "00:49.7",
      "delivery": "Rapid feature listing with payoff timing",
      "uncertain": false
    },
    {
      "end": "00:58.8",
      "line": "Then the app makes a structured complaint-ready record.",
      "start": "00:56.4",
      "delivery": "Concluding system summary",
      "uncertain": false
    },
    {
      "end": "01:05.37",
      "line": "So on one drive to work, if you hit 12 potholes, you have 12 complaints ready by the end of your ride.",
      "start": "00:58.8",
      "delivery": "Satisfying circular resolution",
      "uncertain": false
    }
  ],
  "editing": {
    "cut_count": 28,
    "transitions": "Hard cuts throughout",
    "caption_style": "Single/two-word animated subtitle overlay in bottom center",
    "end_card_or_cta": "None",
    "average_shot_seconds": 2.33,
    "effects_and_overlays": "Split-screen frame, red UI bounding boxes, graphic popups"
  },
  "teardown": {
    "hook": "This Bengaluru engineer built an app that detects potholes, finds the responsible contractor and files a complaint in 4 seconds.",
    "format": "talking-head",
    "mechanism": "utility",
    "structure": [
      "Hook introducing an engineer using AI to automate pothole complaints in 4 seconds",
      "Backstory on Gaurav Sen's commute problem and building the tool with OpenAI",
      "Technical breakdown of hardware and AI vision classification",
      "Civic data layer matching potholes with contractor warranty records",
      "Summary payoff showing automatic complaint generation per commute"
    ],
    "hook_source": "transcript",
    "why_it_worked": "Framed everyday commuter frustration into a high-satisfaction civic tech accountability story."
  },
  "on_screen_text": [
    {
      "end": "00:03.0",
      "kind": "hook",
      "text": "ai just declared WAR ON POTHOLES",
      "start": "00:00.0",
      "style": "Bold white sans-serif uppercase centered overlay",
      "source": "creator",
      "position": "center top"
    },
    {
      "end": "00:06.6",
      "kind": "embedded",
      "text": "Complaint Ready 4 seconds",
      "start": "00:04.7",
      "style": "Green bold heading with round checkmark icon",
      "source": "app_ui",
      "position": "center UI"
    },
    {
      "end": "00:18.1",
      "kind": "label",
      "text": "Codex",
      "start": "00:16.8",
      "style": "Large white sans-serif on blue gradient",
      "source": "creator",
      "position": "center top"
    },
    {
      "end": "00:26.9",
      "kind": "label",
      "text": "Accelerometer ADXL335",
      "start": "00:26.0",
      "style": "Black bold title above circuit board component",
      "source": "creator",
      "position": "center top"
    },
    {
      "end": "00:31.5",
      "kind": "embedded",
      "text": "pothole 0.55",
      "start": "00:29.0",
      "style": "Red bounding box label over detected hole",
      "source": "app_ui",
      "position": "center road"
    },
    {
      "end": "00:33.8",
      "kind": "label",
      "text": "Vision Models",
      "start": "00:31.5",
      "style": "White sans-serif title over dark code screen",
      "source": "creator",
      "position": "top center"
    },
    {
      "end": "00:36.4",
      "kind": "label",
      "text": "Small / Medium",
      "start": "00:35.0",
      "style": "Clean grey category titles over pothole photos",
      "source": "app_ui",
      "position": "center UI"
    },
    {
      "end": "00:56.4",
      "kind": "embedded",
      "text": "Potholes detailed Daily Report",
      "start": "00:45.7",
      "style": "White text on black app header bar",
      "source": "app_ui",
      "position": "top of mobile screen"
    }
  ],
  "structure_beats": [
    {
      "end": "00:06.6",
      "beat": "Core Value Hook",
      "start": "00:00.0",
      "purpose": "State the high-utility promise of instant 4-second contractor complaints."
    },
    {
      "end": "00:22.7",
      "beat": "Commuter Origin Story",
      "start": "00:06.6",
      "purpose": "Establish personal frustration and motivation to build an AI accountability tool."
    },
    {
      "end": "00:36.4",
      "beat": "Hardware & Vision Detection",
      "start": "00:22.7",
      "purpose": "Break down dashcam, accelerometer trigger, and AI vision size classification."
    },
    {
      "end": "00:56.4",
      "beat": "Civic Accountability Hack",
      "start": "00:36.4",
      "purpose": "Reveal warranty legal loophole and database cross-referencing."
    },
    {
      "end": "01:05.37",
      "beat": "Payoff Resolution",
      "start": "00:56.4",
      "purpose": "Deliver circular payoff showing 12 potholes transformed into 12 ready complaints."
    }
  ],
  "recreation_blueprint": {
    "shot_template": [
      {
        "slot": "Hook",
        "visual": "Split screen: top demo of problem detection; bottom creator talking head.",
        "seconds": 6.6,
        "voiceover": "This [Location] [Profession] built an app that solves [Common Pain] in [Short Time].",
        "text_overlay": "[TOPIC] JUST DECLARED WAR ON [PROBLEM]"
      },
      {
        "slot": "Backstory",
        "visual": "B-roll of daily commuter struggle and coding on laptop.",
        "seconds": 16.1,
        "voiceover": "[Founder] suffered from [Pain] daily, so they used [AI Platform] to build a solution.",
        "text_overlay": "[AI Tool Name]"
      },
      {
        "slot": "Tech How-It-Works",
        "visual": "Product hardware close-up and code snippet classification.",
        "seconds": 13.7,
        "voiceover": "Here is how it works: sensor triggers capture, processed by AI vision models.",
        "text_overlay": "[Sensor / Model Label]"
      },
      {
        "slot": "System Secret",
        "visual": "Dashboard UI cross-referencing public legal or government registry records.",
        "seconds": 20,
        "voiceover": "Here is the clever part: matching faults to public records and responsible parties automatically.",
        "text_overlay": "[Database / Metric Name]"
      },
      {
        "slot": "Payoff Outro",
        "visual": "Dashcam driving footage cut to completed report on smartphone.",
        "seconds": 8.97,
        "voiceover": "So on one [Action], you have [Number] automated solutions ready by the time you arrive.",
        "text_overlay": "[Completed Action Status]"
      }
    ],
    "what_can_change": [
      "Target municipal or consumer problem (e.g., parking fines, delayed trains, illegal dumping)",
      "Specific hardware sensors and AI model providers used"
    ],
    "template_summary": "Civic innovation breakdown using split-screen commentary, sensor explanation, and database payoff.",
    "what_must_be_kept": [
      "Split-screen layout pairing talking head with contextual B-roll",
      "Specific metrics (time in seconds, contract counts, daily bumps)",
      "Technical credibility through real code and hardware names"
    ],
    "generation_prompts": [
      {
        "prompt": "Split screen 9:16 vertical video. Top half shows front car dashcam road footage with red bounding boxes tracking road craters. Bottom half shows young male tech creator talking enthusiastically to camera in a modern desk studio.",
        "shot_index": 1,
        "negative_prompt": "blurry, low resolution, saturated cartoon colors, cluttered background"
      },
      {
        "prompt": "Cinematic close-up of compact automotive dashcam mounted on car windshield with futuristic red holographic outline highlighting internal sensors and GPS antenna.",
        "shot_index": 13,
        "negative_prompt": "cracked glass, messy wiring, dark underexposed lighting"
      },
      {
        "prompt": "Clean mobile UI mockup of civic audit dashboard displaying pie charts, zone metrics, and government contract lists on clean modern dark/light mode interface.",
        "shot_index": 25,
        "negative_prompt": "unreadable text, distorted layout, noisy artifacts"
      }
    ]
  },
  "props_apps_and_products": [
    {
      "name": "Qubo Dashcam",
      "how_used": "Hardware device collecting driving video and sensor data",
      "first_seen": "00:22.7"
    },
    {
      "name": "ADXL335 Accelerometer",
      "how_used": "Sensor graphic explaining bump detection",
      "first_seen": "00:26.0"
    },
    {
      "name": "OpenAI Codex / GPT Vision",
      "how_used": "AI models used to analyze footage and build the application",
      "first_seen": "00:16.8"
    },
    {
      "name": "BBMP Pothole Report Portal",
      "how_used": "Public municipal data dashboard queried for road contract data",
      "first_seen": "00:45.7"
    }
  ]
}

model gemini-3.8-flash · analyzed from the actual video and audio · 2026-10-22 · timestamps are estimates

Caption

Gaurav Sen built an AI-powered app that detects potholes, identifies the responsible contractor, and files complaints in just 4 seconds. This smart innovation uses dashcam data and OpenAI’s Codex to turn daily road bumps into instant accountability. #AI #TechInnovation #SmartGadget #UrbanTech #MachineLearning

Is this your account? Ask us to remove @ai.with.andrew: we stop tracking it and delete what we hold about it within a month.

Get this for your own niche

Copy AI founders for free, scan it on your plan, and get teardowns and adaptations for every outlier.

Start with this niche

More outliers in AI founders

Full report