UX × AI: design and ship AI products people trust
Consulting, 1-on-1 coaching and hands-on training for teams building with AI, led by a designer with deep UX/UI roots.
fig. 01Produlogi markglazed ceramic
Consulting for teams building AI products
Compare the four tracks/consultingThe next generation of AI products will act on people's behalf. They will draft the reply, file the claim, match the invoice and suggest the next step. Most of them will succeed or fail at the interface, at the moment a person decides whether to trust what the model did.
AI Products
Pick the AI use cases worth building, design them around real people and ship a prototype you can test. AI product consulting from a UX-first studio./consulting/ai-productsProduct Management
Product management for the AI era: discovery, AI PRDs, evals and roadmaps, as a sprint or a fractional AI PM embedded in your team./consulting/product-managementUX, UI and Product Design
UX research, UI and product design for AI features: interfaces people trust, states that handle uncertainty and research that changes decisions./consulting/ux-product-designAI Automation
Design AI workflow automation people adopt: map the process, prototype the agent and roll it out with the humans in the loop./consulting/ai-automation
1-on-1 coaching for designers, PMs and founders moving into AI products
See the programme/coachingAI product work is becoming its own discipline. Designers are asked to design for uncertainty. PMs are asked to define good output and prove it with evals. Founders are asked to show a working product before anyone will fund a team. The people who make the move over the next two years will be the ones with one real AI artefact to show and the judgment behind it.
| If you are | You leave with |
|---|---|
| Designers | A portfolio-ready AI feature case study, including the research, the AI states and the failure modes, plus a working prototype. |
| PMs | An AI PRD that includes success metrics and an evaluation plan, and a prioritised AI roadmap for a real or realistic product. |
| Founders | A validated AI use case, a working prototype and a build plan or a brief for hiring a builder. |
AI training for the people who shape AI products
See all training/training
Marcus teaches as faculty at AiTraining2U, an HRD Corp registered provider. The programmes run in person in Malaysia and are HRDC claimable through AiTraining2U.
Products we've brought to life
Craft over AI slop
AI makes average output free, so craft is the part people notice. We bring an extensive UX/UI background to that problem, and we keep testing, building and practising how to ship AI experiences that feel designed. Every page and product we ship, this one included, goes through the same five checks.
Run the full slop test01 Visual
A palette someone chose on purpose, and a layout that follows the content.
02 Content
Real data, or sample data labelled as sample. Each number names its source.
03 AI UX
Design the unsure and wrong states, and ask a person before anything irreversible.
04 Copy
Claims with a verb and an object that a reader could check.
05 Process
A real user tries it and someone other than the maker critiques it before release.
Latest lab notes
Read the lab notesOur Own Homepage Failed the Slop Test
Before telling clients to avoid AI slop, we ran our five-row slop test on the old Produlogi homepage. It failed three rows. What we found and changed.
Write the Design Rules Before the Model Guesses
Generated UI drifts to the same defaults because the model fills every gap you leave. A short design rules file, read before each session, closes them.
Design the States the AI Demo Skips
An AI demo shows one good answer. Real users meet the empty, slow, unsure, wrong and corrected states. Put them in the spec before anyone builds.
How we build
Research, design, build and scale: a tight loop where every iteration sharpens your product.
- DiscoverUX research & market insight
- DesignUX, UI & product design
- Build & ShipFull stack execution & launch
- ScaleAI agents, automation & growth
See the loop in detail
- 01 observe
Research the people and the workflow before anyone picks a model.
- 02 frame
Write the spec for every AI state, including what a wrong answer looks like.
spec review - 03 prototype
Build a working prototype with the unsure and error states in place.
- 04 evaluate
Put it in front of real users and label every failure.
fail: back to frame - 05 evaluate
Run the same tasks again after the fixes.
pass - 06 ship
Release with a written handover and evals that keep running after launch.
your approval
Illustrative trace. Span lengths show order, not measured time.
Latest insights
All insights/blog- UX for AI & Product Design
AI Error Handling UX: When the AI Gets It Wrong
Your AI feature will be wrong in front of users. Design recovery first: cheap correction, scoped answers, honest uncertainty and a clean human handoff.
- AI Product Management
AI Product Manager Skills and the FDE Hype
AI product manager skills and the forward deployed engineer role: which skills have evidence, what the hiring data shows, and what mid-size firms need.
- Building AI Products
MVP Development: How to Build the Right Thing Before Building It Right
Most MVPs fail because they test the wrong thing. Learn to scope, design and ship an MVP that tests your riskiest assumption before you invest in scale.
- AI Automation & Adoption
AI Pilot to Production: Scaling User-Centred AI
Most AI pilots stall after the demo. A practical guide to taking an AI pilot to production: baselines, trust, evals, ownership and a 90-day plan.
/internship
UX × AI Product Design Internship
Applications are closed for now. Join the waitlist and we will contact you when the next intake opens.
Marcus Chia
More about Marcus/about
Marcus Chia runs Produlogi, a UX × AI studio based in Kuala Lumpur, Malaysia, working worldwide. He teaches AI courses as AiTraining2U faculty.
Marcus comes from UX and UI design, with an extensive background in designing digital products. He tests, builds and ships AI experiences as an ongoing practice, to learn where they break.
- Base
- Kuala Lumpur, Malaysia
- Hours
- MYT (UTC+8), async first with teams worldwide
Talks & events
Webinar · AiTraining2U · 17 Sept 2026 · Speaker
Building Magical AI Experiences
Webinar · AiTraining2U · 27 Aug 2026 · Speaker
Prototyping via Claude Code for Product Managers
Webinar · AiTraining2U · 23 Jul 2026 · Speaker
Next step
Tell us what you're building
Based in Kuala Lumpur, working with teams worldwide.