For founders and product teams building AI
Creating clear and human-centered AI experiences.
I help teams ship trustworthy AI products by clarifying the model, designing safe interactions, and validating with users.
Diagnose your AI product
Tell me what you're building or struggling with,
you'll get a pattern-based diagnosis specific to your product.
Available on desktop
Open this page on a larger screen to use the AI diagnostic tool.
Grounded in AI UX patterns across transparency, user control, uncertainty handling, and error recovery
Measurable trust
B2B SaaS • 25–30% faster completion
Users got control over AI outputs. Accept, edit, redirect. Confidence followed.
Side Project • High engagement
Financial clarity through conversation, not dashboards. Users stayed longer than any screen-based prototype.
Ad Platform • Validated approach
AI output became a starting point, not a final answer. Users edited instead of starting from scratch.
Where I work with you
You can start with the free diagnostic below to surface what's not working. If you want to go deeper, I offer focused engagements – from a one-time foundations check to a full product partnership.
Foundation Work
Report • v1.0 • 2026-01-30
Onboarding clarity
Users don't understand what the model can/can't do.
Prompt guardrails
Lack of safe defaults + unclear failure states.
Data quality
Input validation + edge cases are covered.
Trust
UX
Data
Current
Target
Prototype onboarding
Week 2
Validate with users
Handoff + roadmap
Week 4
Good AI gets shipped.
Great AI gets used.
The gap is rarely the model. It's the patterns around it: whether users understand what's happening, feel in control, and whether the AI is solving a real problem.
Transparency through conversation
AI shouldn’t be a black box. I design interfaces that explain their reasoning and process.
User Control
Give users agency over AI outputs with explicit actions: undo/redo, edit suggestions, clear state indicators, and a confirm step before finalizing.
Conscious AI decisions
Every feature must solve a real problem, not just exist because it’s possible. Know when to automate and when to empower.
Design for trust, not hype
Building long-term value by prioritizing reliability and user confidence over novelty.
The one in the yellow shirt is me. I'm Aviad :)
With more than five years designing B2B SaaS products, the last few focused on AI. What keeps me in it, honestly, is watching how people make sense of something they can't fully see.
AI keeps changing, but some of the ways users interact with it keep coming back. The hesitation before trusting an output. The moment they realize they can push back. The expectations they arrive with before they even touch the product. Users are building real mental models for AI now, with understanding and expectations to match. I researched these, and the diagnostic tool on this site is built from that work. The products that fit tend to get used.
I work closely with founders and product teams, figuring out together how it should actually behave with users.
Recent articles
UX Patterns for AI Products (That Actually Work)
Designing AI products that users actually trust. Six patterns I keep seeing across different AI products and features.
The part I skipped (and missed)
For the past month I’ve been building my site with Claude Code, Cursor, and a bunch of agents.
Let's talk
Available for freelance and consulting. Berlin-based, remote-friendly.