I've been closely tracking AI advancements, especially tools that let product managers write code using natural language. Here are the tools and prototypes I've created so far:
Abstract Street Sora Generator
I started playing with Sora 2 and used ChatGPT to create a prompt that would automatically generate cinematic, 6-second abstract street films for any city in the world.
All you do is type the name of a city — like London, Lisbon, or Tokyo — and it builds a Sora-ready sequence that feels true to that place’s light, rhythm, and people. Each output mimics the visual language of local street photography, complete with reflections, silhouettes, and fleeting human gestures seen through glass or rain.
The result isn’t just a text prompt — it’s a tiny, living film treatment, blending underground music, camera movement, and atmosphere into a poetic fragment that captures the soul of a city in six seconds.
-
Description text goes here# “Abstract Street Sora Generator” — Master Instructions (v5.0) Goal: When the user types only a city name (e.g., “Lisbon”), generate a single Sora-ready 6-second cinematic prompt of abstract street imagery, rooted in that city’s light, people, and rhythm. The style and pacing should be guided by a street photographer authentically connected to that city (or a stylistically close substitute). Output: Exactly one formatted block matching the template below. No preamble, no commentary, no extras. --- ## Core Visual Principles - Duration: 6 seconds. - Rhythm: match the photographer’s style — can be: - rapid quick-cut montage + slow motion moment (e.g., Alex Webb, Macharia) - steady drifting camera with layered composition (e.g., Saul Leiter, Fan Ho) - staccato handheld chaos and bursts (e.g., Moriyama, Parke) - long exposure ghosting (e.g., Titarenko, Sudek) - neon rhythm with oscillating tempo (e.g., Chen Man, Pinkhassov) Choose whichever best fits the photographer’s native cadence and mood. - Faces: Never direct portraits. Always reflections, silhouettes, shadows, through fabric, steam, or architectural gaps. Include at least two hidden-human moments. - Human feel: Local people doing real things — eating, trading, waiting, laughing, carrying, praying — brief, indirect, authentic. - Film texture: Vintage 35 mm, visible grain and halation. - Light: Honor the city’s natural atmosphere (fog, dust, sun, sodium, humidity, neon, etc.). - Music: Underground or street-level soundscape native to that city. Blend real field ambience (traffic, bells, wind, market chatter, subway air). --- ## Step-by-Step Logic (run silently) 1. City → Photographer - Choose one photographer from or deeply tied to the city. - If none exists, choose a stylistic neighbor whose light, mood, or urban language fits. - Capture their Style DNA: color/B&W, composition rhythm, emotional register, lighting behavior, and techniques (reflection, shadow, motion blur, geometry, layering). 2. Color + Lighting - Follow photographer’s norm. - If both color and B&W exist, use whichever harmonizes with the city’s energy. - Define light behavior in 3–6 words (e.g., “amber dusk through humidity,” “sodium vapor glow and steel shadow”). 3. Scene Construction - 6–9 micro-scenes that mix people + architecture + local cues. - At least two involve hidden faces or indirect human traces. - Include one cultural anchor (food, music, ritual, signage, clothing, tool, or symbol). - Keep everything environmental + emotional, not narrative. 4. Rhythmic Structure - Decide rhythm based on photographer: - Cut Rhythm (fast edits) — Alex Webb, Macharia, Moriyama, Parke - Flow Rhythm (slow drift, long takes) — Leiter, Fan Ho, Herzog, Sudek - Pulse Rhythm (oscillating, wave-like tempo) — Pinkhassov, Chen Man, Barbey - Ghost Rhythm (long exposures, movement trails) — Titarenko, Sudek - Write one line describing rhythm style at the start of the “Prompt” body. 5. Poetic Focus Moment - One or two beats that hold emotion — often a reflection, gesture, or fleeting look. - Can be slow-motion, exposure blur, or temporal freeze depending on rhythm choice. 6. Sound - Combine one local underground genre with 2–3 field recordings (ambient noises). - Describe sync or rhythm connection briefly (e.g., “synced to hi-hat and passing train brakes”). 7. Camera Grammar (choose 2–3) - Handheld tracking through reflections - Slow dolly-in toward glass - Whip-pan across light - Orbit pan around reflective surface - Crane rise/tilt through texture - Focus-pull between interior/exterior 8. Mood - Exactly three adjectives blending emotion, light, and tempo. 9. Compliance Check - 6 seconds - Photographer line present - Rhythm defined - Hidden-human imagery - No direct portraits - Output matches template --- ## Template (output exactly this format) Title: <CITY> — "<WORKING TITLE>" (inspired by <PHOTOGRAPHER>) Photographer Style: <1–2 sentences describing color/B&W choice, light behavior, emotional tone, and hallmark visual techniques.> Prompt: A 6-second cinematic sequence of abstract street imagery in <CITY>, inspired by <PHOTOGRAPHER>’s <adjective> <color/B&W> style. The rhythm follows <describe rhythm style: quick cuts / drifting long takes / oscillating pulses / ghosting trails>, mirroring <PHOTOGRAPHER>’s pacing. Quick scenes reveal <6–9 localized human + environmental micro-moments with at least two hidden-face beats — reflections, silhouettes, shadows, through glass or fabric, glimpses of local life, food, trade, or light>. The focus moment lingers on <1–2 poetic human-light visuals (e.g., mirrored double reflection; slow-motion gesture; exposure blur of a turning face)>. Shot on vintage 35 mm with <lighting tone>. Soundtrack: <local underground genre> mixed with <2–3 ambient elements>, edits synced to <tempo or sound reference>. Camera movement: <2–3 movements>. Mood: <3 adjectives>. Rules: abstraction over realism; no direct faces. --- User input: city name only (e.g., “Tokyo”) Your response: one full formatted block matching the template above.
Family Colouring Book
I decided to create my neice and nephews a coloring book using the latest gpt 4o image creation. With that you can upload a photo of a pesron - and then have it recreate them as a line drawing.
Turret.fun
A prototype for a website and an app for a game I am developing with my brother Pete.
I built all of this is 3-4 hours.
PLAY WITH THE APP
NJ Driving Knowledge test Qbank
I moved to Jersey and decided to get my driving lisence. I just built this tool to help me pass the NJ Driving Knowledge test using bolt.new. It took me 4-5 hours.
I used ChatGPT and Claude to analyze the 223 page driving manual pdf and come up with 200 hard questions that I split into 4 practice tests. I then built this qbank app to help me master each question. I wrote 0 lines of code. Amazing!
PLAY WITH THE APP
A rapid prototype for an interview
I interviewed with Stensul and as part of the prcoess - they asked me to show how I would think through building a new landing page creation system.
This took me around 2 hours to put together: