HumanBuiltAI
Scott Galetto/Portfolio/2026

AI-fluent design technologist.

I help teams figure out what to build, then bring it to life through design, prototyping, and AI.

Selected work

Selected work

06 case studies
Specs cover: art director glasses with brand assets in one lens and specs for AI in the other, over the line A brand kit for AI
01 · 2026 AI product · Brand systems
Specs A brand kit for AI. My own app turns a small business’s messy logo, PDFs and photos into a kit AI can follow, then art-directs on-brand posts that explain their choices. Read case study
Aon / Horizon cover: a climate risk grid with an “Explore the risk” callout, Aon logo lower right
02 · 2018 Product design · Data viz · AI
Aon / Horizon Complex data, made explorable. A climate-risk visualizer rebuilt around the questions people bring to it, with a conversational layer for asking the data. Read case study
Chipotle / RAINCHECK cover: a raincheck ticket reading “One free burrito. Text, Wallet, enjoy.”, Chipotle logo lower right
03 · 2016 SMS · Mobile Wallet · Animation
Chipotle / RAINCHECK A closed door, turned into a free burrito. The text-to-Wallet campaign that drew 5.3 million requests in about five hours. Read case study
CreditNinja cover: a single question card, “How much do you need? $1,500”, with a slider, CreditNinja logo lower right
04 · 2021 UX · Conversational form
CreditNinja A 15-minute form, now under five. A loan eligibility flow redesigned as a conversation, with a 40% lift in completion. Read case study
Sauza / Lasso Up With A Cowboy cover: a lasso loop around “One throw. Two screens.”, Sauza logo lower right
05 · 2014 Interactive kiosk · Game
Sauza / Lasso Up With A Cowboy One throw, two screens. A phone-to-iPad in-store game that won a Verizon Partner Innovation Award for its use of WebSockets. Read case study
Verizon / IndyCar cover: a race track outline with “Race day. In your hands.”, Verizon logo lower right
06 · 2014 Mobile web · VR
Verizon / IndyCar Race day, in fans’ hands. A mobile web and VR activation that lifted engagement with Ways to Win by 124% over the prior year. Read case study

From the studio

See the studio

Brand identities and live code experiments: everything that doesn’t fit in a case study.

Specs cover: art director glasses with brand assets in one lens and specs for AI in the other App · Claude APISpecs Six Degrees of Kevin Bacon, dressed as a 1996 homepage App · TMDB APISix Degrees of Kevin Bacon The Peacock Room identity, built on a fanned peacock-feather mark BrandingThe Peacock Room Minibar logo BrandingMinibar Jim Cachey Realty identity: a fine-line skyline inside an arched window BrandingJim Cachey Realty Brockway Photography identity with a quiet serif wordmark BrandingBrockway Photography A 3D miniature of Wicker Park at night: brick buildings, wet streets and a Blue Line train pulling into Damen station Lab · Three.jsWicker Park after the rain A 3D miniature of an East Village block with a small park Lab · Three.jsEast Village miniature Painted WebGL flowers growing up from the bottom of the canvas Lab · WebGLWebGL flowers HumanBuiltAI spelled in colorful animated SVG letters Lab · GSAPHumanBuiltAI letters A field of dots rippling outward Lab · WebGLRipple dot field A fan of tilted photo cards Lab · CSSScroll-driven tilted cards A small monster drumming at a laptop Lab · SVGLittle Monster Jams Neon line drawing of the twins standing in a hotel corridor under the words Come play with us Video · RemotionThe Shining homage

Approach

AI drafts.People make it great.

Speed is half the value. Judgment is the other half: knowing what a good result has to do before asking for one, and checking every draft against it. That balance will shift as the models get better. Knowing when to let go is part of the job.

InputsMessy, repetitive work
AI connects to your businessLearns your data, brand assets, style and tone, then writes the first draft
Human judgmentReviews, edits, adds context, brings creativity and empathy
A better resultApproved. Polished. More human.
01Inputs
Messy, repetitive work
02AI connects to your business
Learns your data, brand assets, style and tone, then writes the first draft
03Human judgment
Reviews, edits, adds context, brings creativity and empathy
04A better result
Approved. Polished. More human.
01Define good firstBefore I prompt anything, I write down what a successful result has to do, who it’s for and what it can’t get wrong.
02Ask for optionsI ask AI for alternatives, not one finished answer, so there’s something real to compare and push against.
03Review against the goalEvery draft is judged on whether it’s useful and accurate, not how fast it arrived. What I correct sharpens the next brief.

The brand system is the brief

  1. 01One prompt
  2. 02Brand files for people
  3. 03A system for agents
  4. 04On brand by default

The prompt

“Design the ‘order ready’ screen for our grocery pickup app.”

Same prompt, same AI. The only thing that changes is what the agent can read.

What the agent can read

brand/for-people/Unstructured

Brand-Guidelines.pdfLaunch-Deck.pdfPhoto-Library/

Made for people to browse. An agent can skim it, but it has to guess the rest, so the first draft comes out generic.

brand/for-agents/Structured

README.md tells the agent which file to open for which job.

Turn on each file to see the screen change.

Exact values, not a description of them. Fieldstone green is #2F6B4F, harvest gold is #E8A33D, corners are 14px, Newsreader sets the headlines and Geist does the rest. There’s nothing left to interpret.

Every rule carries its reason. Pickup is a promise, not a notification, so the screen says where and until when instead of “Your order is ready!”

Rules say where they stop. Service screens carry no promos and no ratings, so the agent can’t trade someone’s errand for an upsell.

Real parts, not lookalikes. The pickup code, the swap approval and the I’m-here bar come straight from the app’s React components.

Same prompt, same AI. Four files made the first draft look, sound and behave like Fieldstone. Set the system up once and every agent starts on brand. Specs does that setup for small businesses: it reads their messy files, builds the kit, then makes on-brand posts from it. Try Specs

Built a working version of this: Specs

Toolkit

The toolkit, in daily use.

Design, AI, code and automation, by discipline and by how often I reach for them. AI now sits at the center of my day, next to design tools that have been there all along. One layer per tool, and the thicker the layer, the more I use it. Select a stack to see what’s in it.

Select a discipline ↙One layer = one tool
In the stack

Daily · Regular · Familiar
    Layer thickness · Daily / Regular / FamiliarTap a stack to explore
    About · The person behind the work
    Portrait of Scott Galetto
    Scott GalettoChicago, Illinois

    A designer whobuilds the thing.

    Print to digital, desktop to mobile, native apps to the web, and now AI. Every shift asked the same thing: figure out what the new technology makes possible, then make it useful.

    I’ve done that from more than one seat. Art director and developer at Vibes. UX lead in Aon’s technology incubator. Creative director at CreditNinja, where a conversational loan form became a real product. Agency and in-house, design and code. It doesn’t fit one box, and that’s the advantage.

    Today I build GenAI apps, prototypes and workflows with Claude and Codex. Code was the first craft to stop being typed by hand. Design and marketing are next, and I help teams make that move with their judgment intact.

    The full story / Download résumé
    1. 2024–NowHumanBuiltAIIndependent practice: AI products and prototypes
    2. 2022–2024Audio Video DynamicsSenior Product Designer
    3. 2020–2022CreditNinjaCreative Director
    4. 2017–2020AonSenior Front-end Developer, Experience Designer
    5. 2010–2017VibesArt Director, Web Developer, Senior Experience Designer
    Contact

    Need to make it make sense? Let’s talk.

    Message didn’t send. Please try again in a moment.
    Thanks. I’ll be in touch.
    HumanBuiltAI
    © 2026 Scott GalettoLinkedIn ↗Back to top ↑