CARLOS GARAVITO
MENU
FITPROOFSYSTEMRESUMELET'S TALK
FOUNDING PRODUCT DESIGNER / AI-NATIVE

CarlosGaravito

FOUNDING PRODUCT DESIGNERAI-NATIVEDESIGN + PRODUCT + CODE

I design the product, build the system, and use AI to close the distance between idea and shipped software.

Designer and creative technology leader with 20+ years shaping digital experiences, products and systems. Today I work AI-natively across product definition, UX, prototyping, coded design systems and generative production.

SEE THE FITRESUME ↗
TORONTO / HYBRID READYDESIGNERPRODUCT THINKERAI-NATIVE BUILDERSYSTEMS THINKER
02 / THE FIT

SELLIT9 is designing for two customers.
People and agents.

That is the part of this role I find most compelling. The next generation of commerce will not live only in pages and funnels. Products need to work through interfaces, APIs and AI agents while preserving trust, brand and conversion logic.

PRODUCT

Frame problems, define requirements, prototype, prioritize and ship.

DESIGN

UX, interaction, visual systems, conversation design and brand.

AI

Codex, Claude, agentic workflows, prompt/context systems and production automation.

SYSTEMS

Reusable rules, components, governance, QA and capability transfer.

03 / SELECTED PROOF

Four proofs.
One operating model.

Each case proves a different part of the role. Nothing here exists simply because it is impressive.

01
DESIGN SYSTEM / AGENT-READY RULES

cgOS

SYSTEM SPECIMEN / TOKENS
--accent#FF4FD8--rule1px / hardtypedisplay + mono

I turned my own digital identity into a governed design and experience system.

PROOF

Tokenized typography, reusable components, interaction rules, responsive standards, content principles and implementation guidance designed to reduce one-off decisions across multiple properties.

WHY IT MATTERS HERE

Direct evidence for maintaining brand and design rules as a system that humans and coding agents can implement.

02
AI-NATIVE PRODUCTION SYSTEM

ENGINE_v1.0

WORKFLOW FRAGMENT / ENGINE_v1.0
BRIEF→CONTEXT→GENERATE→QA→APPROVE

I build repeatable AI workflows, not isolated prompts.

PROOF

A governed production architecture moving from brief through brand system, creative direction, generation, variation, QA, approval, measurement and learning.

WHY IT MATTERS HERE

Evidence of reusable context, QA gates, agentic workflows and system-level thinking.

03
PRODUCT + OPERATIONAL IMPACT

3 WEEKS → 24 HOURS

I changed the workflow, not just the output.

PROOF

At ASTOUND, I conceived an AI-assisted system that transformed structured experiential briefs and constraints into configurable design directions, reducing an RFP response workflow from approximately three weeks to 24 hours.

WHY IT MATTERS HERE

Evidence of product framing, workflow redesign, implementation thinking and measurable operational impact.

04
AI PRODUCT EXPERIENCE

conversationHEALTH

Designing conversations before agents were mainstream.

PROOF

Led experience strategy, UI/UX, conversation design and creative direction for AI/ML-powered healthcare voice assistants.

WHY IT MATTERS HERE

Direct evidence designing natural-language product interactions where trust, clarity, user intent and system behavior must work together.

04 / DESIGN IN CODE

The design does not stop at the mockup.

My current workflow moves directly between product definition, design, code and QA. I use coding agents to turn decisions into working software, then evaluate the implementation in the browser rather than treating handoff as the finish line.

PROBLEM→SPEC→FLOW→CODED PROTOTYPE→BROWSER QA→ITERATE→SHIP→MEASURE

AI accelerates implementation. Judgment remains human-owned.

05 / AGENT-READY DESIGN

A design system should be executable.

The interesting shift is not simply Figma to code. It is human-readable design intent becoming machine-operable context.

HUMAN LAYER
  • Principles
  • Visual language
  • Interaction
  • Content
  • Accessibility
ONE
SYSTEM
AGENT LAYER
  • Tokens
  • Component contracts
  • Allowed variants
  • Constraints
  • QA gates
  • Implementation instructions
06 / PRODUCT EXPLORATION

What happens when the customer brings the interface?

An independent exploration of agent-native trade-in. Natural language becomes structured state, then only the missing information is requested before a service handoff.

INDEPENDENT CONCEPT / NOT A SELLIT9 PRODUCT

CUSTOMER

I have a 2023 MacBook Pro M2, good condition, slight scratch on the lid. What can I get for it?

AGENT

I’ve turned that into structured trade-in details. Review what I understood and correct anything before continuing.

STRUCTURED STATE

07 / PRODUCT JUDGMENT

Design decisions need a metric.

MEASUREMENT PLAN / NOT SELLIT9 DATA

01START QUOTE
02DEVICE IDENTIFIED
03CONDITION COMPLETED
04QUOTE VIEWED
05CHECKOUT STARTED
06TRADE CONFIRMED
GA4

Funnel completion, drop-off, source and device.

CLARITY

Rage clicks, dead clicks, hesitation and replay patterns.

RESEARCH

Comprehension, trust and error recovery.

EXPERIMENT

One hypothesis at a time. Example: reducing the questions shown before device identification should improve quote-start completion.

08 / EXPERIENCE

Senior enough to frame the problem.
Hands-on enough to make the thing.

2022—NOWCommandAlpha_Studios

Founder & Executive Director, AI Strategy

2025—2026ASTOUND

Director, Creative Innovation

2021—2022VML / Ford Canada

VP, Executive Creative Director

2019—2020conversationHEALTH

VP, Chief Experience Officer

2018—2019Deloitte

VP, Creative Strategy

09 / WHY SELLIT9

The interface is becoming a participant.

SELLIT9 sits at a useful intersection: commerce, recommerce, merchant infrastructure and AI-agent interaction. The Founding Product Designer role is compelling because the design problem is larger than screens. It includes the rules, systems and context that allow a product to remain coherent when customers, merchants and agents all touch the same journey.

That is the kind of design problem I want to work on.

10 / LET'S TALK

Let's
build it.

CARLOS GARAVITOToronto, Canada
EMAILLINKEDIN ↗RESUME ↗