Research
The questions behind the work. These are the threads we keep pulling on — the ones that shape how we design and build every product, and everything we publish.
What we believe
A product that isn't understood fails, however smart it is. We design for understanding before aesthetics — and before novelty.
The model is a commodity; the decisions around it aren't. What the AI does, and where, matters more than what it can do.
Intelligence belongs in the flow of the work, not in a tab off to the side. Context is what turns AI from a detour into leverage.
AI is confidently wrong sometimes. Products that plan for that — visible, editable, reversible — earn trust; ones that assume it's right lose it.
People skip instructions, act on intuition, and expect speed. We design for how they actually behave, not how a demo assumes they will.
We reduce noise, not capability. Making the right thing obvious is harder than adding another feature — and worth more.
Threads we're pulling on
How much should an AI show its work? We study the patterns that let people understand, verify, and stay in control of AI output — without drowning them in detail. Trust, we keep finding, is a design property, not a model property.
Read on this
The move from AI-as-a-tab to AI-in-the-work. Where intelligence belongs in a workflow, how it should ask permission, and when it should stay quiet. The bolt-on chatbot is easy to ship and easy to abandon; the interesting work is weaving AI into the grain of the task.
Read on this
Most AI products try to do everything. We research how to find the single loop that proves the product — and scope a buildable MVP around it. Almost every stalled build we see traces back to a loop that was never actually decided.
Read on this
As models commoditize, the edge moves to product judgment: deciding what the AI should do, and where, matters more than what it can do. Two teams with the same model build wildly different products — the gap is judgment, and we think that's where the moat is moving.
Read on this
Research notes
A pattern we keep seeing: AI features get abandoned not because they're wrong, but because users can't tell when they're right. Notes on the trust gap and how design closes it.
Read the note →Zenith Studio · Jun 26, 2026 · 6 minAs models commoditize, the defensible edge moves from the technology to the decisions around it. Notes on why product judgment — not the model — is becoming the moat.
Read the note →Zenith Studio · Jun 18, 2026 · 5 minThe bolt-on chatbot ships fast and demos well, then goes unused. A note on why AI belongs in the flow of the work — and what changes when you move it there.
Read the note →Zenith Studio · Jun 12, 2026 · 5 minEvery product that works has one loop users return to. A note on why naming it is the highest-leverage decision in an AI MVP — and why skipping it is why builds stall.
Read the note →Zenith Studio · Jun 6, 2026 · 5 minAI lowered the cost of building, which means more products get built that nobody needed. A note on painkillers, vitamins, and why demand is the only test that matters.
Read the note →
Zenith AI.
Product studio assistant
Quick answers about Zenith — our services, our process, and where to start with your AI product.
Strategy, experience design, and AI engineering in one senior team — idea to launch.
Define → Design → Build → Improve. We scope tight, build in the open, and you own it all.
Start with a question
