- Owns
- Scope, roadmap, what gets built next
- Shipped
- Killed features, held the launch date, blocked scope creep
Product & UX, at AI speed
I'm Ofir, an AI-native
product designer.
I move at a pace
teams don't expect.
Twelve years of product and UX judgment, five of them leading product design inside cybersecurity. What's different now is speed: decisions become flows, specs, and working prototypes in days instead of months.
The work
What you actually get
Two things, combined. Neither one works on its own.
Senior product and UX judgment
Finding the real gaps between the product, its users, and the business goal. Knowing what to cut, what to fix first, and what doesn't matter yet. This is the part AI cannot do for you.
AI-accelerated execution
Those decisions turn into flows, specs, and a working, demoable prototype fast enough to change what you decide next. Days, not months. Real artifacts, not slides.
The AI setup I build to work with you is yours to keep. It's how I go fast, not what I'm selling.
Background
Twelve years in product design. Five of them inside cybersecurity.
Where I've done it
-
Axonius
CybersecurityDirector of Product Design Feb 2021 - Apr 2026
Built the product design function from zero, joining as Head of Product Design. Led the design team and the front-end team.
-
Salesforce
Senior Manager, UX/UI Jan 2020 - Feb 2021
Led the Israeli Service Cloud design team.
-
ClickSoftware
UX Lead Jul 2017 - Jan 2020
Established the UX and UI practice from scratch.
-
Vizrt
UX Designer Nov 2015 - Jul 2017
Who it's for
Three ways in
Different starting points, same sprint. Pick the one that sounds like you.
Starting
"I want to build something, but I'm not sure how to go about it."
A real idea, no product yet. I define what it actually is, then set you up to build it and stay alongside you while you do.
Start at Shape
Built
"I have something, and I need a review."
An MVP or a shipped feature that isn't clear or launch-ready yet. What's missing isn't more engineering, it's judgment.
Start at Assess your product
Scaling
"I'm already building with AI, and I want it to hold up."
It works, but it's ad hoc. Results vary and nothing carries over between sessions. I put structure around how you already work, so it keeps producing.
Start at Assess your setup
The sprint
Assess, Shape, Build
One sequence, about six weeks end to end. Stages can also be taken on their own.
-
01
Assess
Product Gap Review · ~1 week
An audit against your business goals: user flows, onboarding, information architecture, launch readiness. You get a prioritized gap map with annotated screenshots, severity and impact per issue, and clear next steps. Not a deck. A working document you can act on the same day.
-
02
Shape
Product Definition · ~2 weeks
Findings become decisions. Flows, spec, requirements, scope, design-system fit. A Product Blueprint ready to hand to whoever builds it: your team, a freelancer, or me.
-
03
Build
AI-Assisted Prototype · ~3 weeks
A working, testable, demoable prototype built with AI tools and your existing components. Production-minded prototyping: fast, but built to be believed, not a mockup.
Proof
I did it on my own product first
Glimps is a UX inspiration gallery for the cybersecurity industry. I built it solo, and it's live.
- Companies
- 117
- Real product screens
- 3,500+
- Signups in week one
- 77
I didn't build it alone in the usual sense. I direct a team of AI agents, each owning a domain, each with persistent memory, each with the authority to push back on me. This is the same setup I stand up for clients.
The system that shipped Glimps
One operator · five agents
The product
Glimps
Live in production
- Owns
- Design system, component library, visual consistency
- Shipped
- Design tokens, Storybook, shared components across the app
- Owns
- Sourcing, extraction, quality standards
- Shipped
- 3,500+ screens sourced, named, and tagged
- Owns
- Positioning, launch, retention
- Shipped
- Launch sequence, week-one post, lifecycle email
- Owns
- Architecture, feasibility, infrastructure
- Shipped
- Auth, delivery pipeline, integrations
Each agent holds its own memory, its own standards, and the authority to tell me no. I make the calls they escalate.
This works, and it has real limits. I know which decisions still need a human because I found out on my own product, not on a client's.
Next
Let's talk
A short conversation to see whether there's a real gap worth a Sprint. No pitch deck, no automated report. If there's nothing public to look at yet, a short screen-share works just as well.