Create a vertical 2:3 character reference card for [Luna Solace] as a clearly adult real-human cosplay subject, 17-year-old futuristic female tennis prodigy. Multi-panel layout: bold title, front close-up, confident front stand one hand on hip, back-facing stand, over-shoulder hand-in-hair-nape, seated hugging raised shins, detail boxes, palette row.
Character description: Stylish athletic 17-year-old girl with sharp confident eyes, slim toned build, slightly messy black hair in a sporty high bun with loose strands, fashionable tennis visor/headband, cool-girl competitive expression. She has elegant yet powerful posture.
Outfit Version 1: Oversized white cropped tennis shirt, neon yellow athletic shorts, neon yellow visor, chunky white/yellow sneakers, wristbands, neon yellow tennis racket.
Outfit Version 2: Oversized charcoal athletic tee, cream tennis shorts, mint green headband, mint-accent sneakers, mint grip racket.
STYLE: 90s/Y2K film-flash snapshot editorial mixed with premium anime x fashion hybrid, point-and-shoot disposable camera look, hard direct on-camera flash, cold blue film scan color cast, visible grain, slight motion blur, imperfect tilted framing, paparazzi-style raw fashion snapshot, late-night dressing-room atmosphere, matte skin with tiny natural imperfections, not studio-polished. Cel-shaded semi-realistic rendering with glossy fabric textures on sportswear, stylized proportions with realistic anatomy, vibrant cyan, neon yellow, cream, and charcoal palette.
Preserve strong character recognizability with high-fashion athletic concept art details, modern street-sport aesthetic, energetic and confident mood. Same identity and facial likeness across all panels. Realistic fabrics and materials (breathable athletic mesh, technical fabrics, rubber soles).
Include subtle modern esports/tennis campaign elements: small "VS" graphic, player profile info (Name: Luna Solace, Age: 17, Height: 168cm, Play Style: Aggressive All-Rounder), color swatches, and clean graphic accents.
Mood keywords: competitive, stylish, sporty, youthful, fashion-forward, elite tennis prodigy, confident, energetic, tournament ready.
Negative prompt: low quality, blurry, bad anatomy, extra limbs, deformed hands, realistic photo only, overly cartoonish chibi, old-fashioned design, flat lighting, messy composition, dull colors, generic sportswear, heavy cyberpunk, low detail, pixelated, text errors, plastic skin, studio perfection.
Luna Solace. Tennis prodigy. 90s disposable camera aesthetic meets cel-shaded fashion editorial. Character ref card — two fits, full palette, six panels.
Prompt below
8/ and if you see viral claims about specific capability percentages or job replacement numbers floating around — treat them as anecdotal unless they come with primary docs or repeatable tests. the hype cycle is loud right now. stay skeptical, stay curious.
7/ the takeaway isn't "one lab did something bad." it's that our main tool for checking whether AI is safe — evaluation — is getting gamed (intentionally or not), while international coordination on safety is fraying at the same time. not a great combo.
The 'pretty vs effective' gap usually starts in discovery. Client didn't know what they needed, chose vibes over outcomes. I built a tool that surfaces this early—turns messy client conversations into specs that force clarity on what success actually looks like. Would've saved you both six months.
Some themes I'm watching & predictions for 2026:
1/ AI ads in major chatbots
The economics of free LLM inference are brutal. Expect major providers to experiment with ads/affiliate links. Google's already testing this in AI Overviews.
2/ Proactive AI goes mainstream
AI that works *without* being prompted. Background research, monitoring, drafting. The tech exists—2026 is when the UX catches up.
3/ Agentic interfaces for knowledge work
A "Claude Code for everything" emerges. Persistent context, file access, multi-step execution. Someone cracks the abstraction.
4/ Computer use becomes real
Models reliably operating browsers and apps. Not perfect, but good enough for real workflows. The API-less future begins.
5/ Continual learning breakthrough
One of the labs announces models that actually update knowledge without forgetting everything else. The holy grail gets closer.
6/ A $100B+ AI company files to IPO
OpenAI? Databricks? Market conditions permitting, someone pulls the trigger.
What am I missing?
People say AI use is still cognitive work. You prompt. You evaluate. You synthesize.
True. But that work is front-loaded and meta-level. The middle gets handed off—the slow grinding through, the wrong turns, the building of understanding through effort.
The middle might be where something important happens. Not because struggle is virtuous. But because getting lost is how you learn the geography.
You can check an AI's answer for coherence. You can't check for what's missing—not if you never walked the territory yourself.
@SuhailKakar Great article.Are you finding any approaches that help with the "brief writing" phase? Like frameworks, or ways to stress-test your specs before you start building?
Here is the prompt try it out and repost examples:
metadata:
prompt_name: "Coca-Cola Frosty Reveal"
base_style: "cinematic, photorealistic, 4K, high dynamic range"
aspect_ratio: "16:9"
scene_description: "A vast, dry, sun-baked desert landscape under a harsh midday sky. Cracked earth stretches to the horizon. Heat shimmers rise from the ground. Utterly still and silent."
camera_setup: "A single, fixed, wide-angle shot from low angle, capturing the horizon and the center of the frame where the action unfolds. The camera does not move for the entire 8-second duration."
key_elements:
- "A single, unopened glass Coca-Cola bottle with condensation forming"
- "Iconic red and white Coca-Cola logo clearly visible"
- "Frost radiating outward across the desert"
negative_prompts: ["no people", "no text overlays", "no modern structures", "no animals", "no distracting music"]
timeline:
- sequence: 1
timestamp: "00:00-00:01"
action: "In the center of the scorching desert, the sand begins to tremble. A faint vibration pulses outward in a perfect circle."
audio: "Deep, subsonic hum. The silence of the desert amplifies the tension."
- sequence: 2
timestamp: "00:01-00:02"
action: "With a soft *crack*, the earth splits open. A glowing blue light surges from below. A glass Coca-Cola bottle rises slowly from beneath the sand, pristine and ice-cold, droplets of condensation forming instantly on its surface."
audio: "A crystalline *zing* as frost forms. A low *whoosh* of cool air breaking through heat."
- sequence: 3
timestamp: "00:02-00:06"
action: "Hyper-lapse effect: From the fixed wide shot, frost explodes outward across the desert in a perfect wave. The cracked earth greens with life — grass, wildflowers, and palm saplings burst through. A shaded oasis forms around the bottle: a wooden bench, a straw sunhat hanging on a branch, a cooler nearby. Everything appears as if summoned by the bottle’s presence."
audio: "Rising, melodic chimes layered with ASMR-like sounds: ice clinking, water droplets, soft rustling leaves, distant birdsong emerging."
- sequence: 4
timestamp: "00:06-00:08"
action: "The final droplet of condensation rolls down the bottle. The oasis settles into stillness. Sunlight glints perfectly off the red label. The words 'Taste the Feeling' shimmer subtly in the heat haze — not as text, but as a natural ripple in the air."
audio: "The crisp, iconic *fizz* of a Coke bottle opening — then cut to silence. A single refreshing sip sound. Peace. Coolness. Satisfaction."
This Coca-Cola ad was NOT made by humans. This is Google's Veo 3 AI. Look at that seamless desert → oasis transformation. The bottle stays PERFECT while the entire world changes around it. The condensation. The lighting. The text appearing on the label. VFX studios should be terrified. 🤯 #Veo3 #AIVideo #Advertising #Qwen
Used new Qwen3-235B-A22B-2507 to create the prompt. Complete insane!
Prompt below