RAMSDEN MADEUS / AI CREATIVE SPECIALIST

AI CREATIVE — SHORT-FORM VIDEO / UGC PIPELINES

5 INFLUENCERS 2 UGC ADS 3 REELS 3 HOURS

Ramsden Madeus — I create short-form video with AI end to end: script, face, voice, motion, edit. Ten assets, one person, one sitting. AI isn't the pitch. The turnaround is.

Watch the ads

2 UGC ADS

1.5 HOUR

3 AI INFLUENCERS + REELS

1.5 HOUR

12 TOOLS

IN ROTATION

POV

Volume is the strategy.

01

Speed compounds

If a concept takes an hour instead of a week, you can be wrong forty times and still finish ahead.

02

I keep the stack open

Something new is released every month. I test it myself, drop what doesn't work, and keep what makes my workflow more efficient.

03

Taste still decides

AI only generates options. Having an eye to pick the one that stops a thumb is the job.

The work

UGC ADS

HIGH-END ARABIAN COLOGNE

Product Reaction
“SMELL LIKE A LEGEND”
BUILD TIME
~30 MIN
FORMAT
9:16
TOOLS USED
4 TOOLS
Nano BananaArcadsClaudeClaude

DR. DENT TEETH WHITENING STRIPS

PRODUCT REVIEW
“JUST WIPE, PEEL, STRETCH, AND STICK”
BUILD TIME
~1 HOUR
FORMAT
9:16
TOOLS USED
6 TOOLS
HiggsfieldArcadsElevenLabsNano BananaCapCutClaude

Talent

Reusable cast.

Three AI creators with consistent faces and voices, built in about an hour — so a brand can keep testing with consistency.

Rachel

STREETWEAR · EFFORTLESS COOL

Hailey

FITNESS · WARM & GROUNDED

Tatum

GYM REGULAR · HIGH ENERGY

Process

How it gets made.

  1. 01 · WRITE

    Concept, hook variations, and script. I use Claude to pressure-test prompts and copy, not to write the final line for me.

    Claude

  2. 02 · BUILD THE CHARACTER

    Face, wardrobe, and setting, kept consistent across shots.

    HiggsfieldNano Banana

  3. 03 · GENERATE MOTION

    Performance and camera movement, shot by shot.

    HiggsfieldSeedance 2.5ArcadsRunwayGemini Flash

  4. 04 · VOICE

    Voiceover and delivery direction.

    ElevenLabs

  5. 05 · FINISH

    Upscale, cleanup, cut, captions.

    TopazCapCutOpusClip

Stack

What I'm currently using.

Last rebuild 07/2026.

Claude

CORE

Prompt engineering and copy. Hook variations, script drafts, structuring shot lists, and stress-testing my own ideas before I burn generations on them.

Higgsfield

CORE

Where most of the work happens. Character creation, image-to-video, and motion control; my default for getting a believable performance out of a still.

Seedance 2.5

CORE

Realistic multi-shot generation. My pick when I need natural body movement and dialogue that holds up in close-up.

Arcads

CORE

Where the ad gets built.  My primary tool for getting believable performance and camera movement out of a still - with the product in frame and/or in someones hands.

Nano Banana

CORE

Reference-based image editing and character consistency. Same face, new outfit, new location, without redesigning the person.

Topaz

CORE

Finishing. Upscaling to clean 1080x1920, sharpening detail, and fixing the softness and artifacts that AI video leaves behind.

CapCut

CORE

Cuts, captions, and final assembly. Fast timeline editing and auto-captions for getting a build export-ready without leaving the browser.

Foreplay

EXPLORING

Ad research and swipe library. I'm early with it, but organizing competitor ads into briefable, taggable references is the missing input in my process and I'm building it into my daily workflow now.

Runway

SUPPORTING

Generation and in-editor fixes: extending shots, inpainting, and cleaning up frames that almost worked.

Gemini Flash

SUPPORTING

Fast multimodal passes. Quick concepting, reviewing frames, and iterating on copy when speed matters more than polish.

OpusClip

SUPPORTING

Cutdowns and captions. Reframing, auto-captioning, and spinning one build into multiple lengths.

ElevenLabs

SUPPORTING

Voice. Voiceover generation, cloning, and delivery direction — pacing, emphasis, and tone, which is usually what makes a read sound human instead of synthetic.

What's next

Where I want to point this.

I started creating AI content in July. I was doing so at the same time as being enrolled in a full-time graduate program requiring at least 40 hours a week, and I worked another 20 hours a week managing the social media for a local gym. There was no room for it, but I still found the time. Whether it was through waking up early in the morning, in between classes/rehearsals, or going to sleep late, I made it happen with nobody to hold me accountable, no deadline, and no financial compensation. Can you call me obsessed? You'd have a strong case. But I'd reframe this way: this field is 18 months old, and whoever is good at it in 5 years is the person doing reps now. I'm driven by the fact that I believe AI is the future of content, and I'll do what it takes to be at the forefront of this shifting industry. I grew up as a multi-sport athlete, starting with football at 8 years old, and playing many sports along the way, such as soccer, basketball, and running track & field at a collegiate level. I'm grateful for how sports have taught me about having a Mamba Mentality. (RIP Kobe) Having an athlete's mindset has taught me the importance of putting reps in, overcoming challenges, and having the discipline to show up every day and figure out how to improve my game. That's the mentality I bring to anything that I take on. Whether it's tennis (my current sport obsession), acting, or creating short form/long form content. The work may look different, but the approach is the same. I'll be done with both grad school and my job managing social media in September, and moving back to the US. I'm very excited about this new chapter where I'll be able to shift my focus full-time into AI content creation.