About Manish
A systems thinker who happens to design interfaces.
I design decision systems that help people and AI solve problems together. The interface matters, but it is the last expression of the work, not its center.
Current practice
I am a UX consultant and agentic designer at Deloitte USI. I work where enterprise software meets machine intelligence: regulated products where AI can move fast, but a person must remain accountable.
I am strongest in the 0 to 1 space. I enter when the roadmap is unclear, the workflow is tangled, and the product has not found its shape. I map the system, decide where responsibility belongs, and build enough of the product for people to react to reality instead of a presentation.
I prototype in code as well as Figma. The handoff I aim for is a working artifact that carries behavior, context, and design rules, not a folder of screens a developer has to reverse-engineer.
What I bring
Product judgment with technical range.
01
Decision architecture
I map actors, incentives, rules, data, and consequences before reducing the problem to screens.
02
AI product design
I draw explicit boundaries between probabilistic assistance, deterministic logic, and human judgment.
03
Design engineering
I build functional prototypes in code and preserve intent through systems, tokens, reviews, and handoff.
04
Practice enablement
I turn repeatable work into skills, harnesses, workshops, and shared operating language.
Core belief
AI lowered the floor for building. It did not lower the bar for knowing what is worth building.
As execution becomes cheaper, framing, logic, taste, and judgment become more valuable. I use AI to accelerate implementation, explore alternatives, and inspect edge cases. I do not use it to make consequential decisions disappear.
How I work
Sense. Model. Decide. Build. Systemize.
Sense
Gather evidence, challenge assumptions, and map the current reality.
Model
Make actors, constraints, information flow, and open questions explicit.
Decide
Set the product logic and the boundary between AI and people.
Build
Use a working prototype to replace imagined feedback with observed behavior.
Systemize
Turn what worked into a reusable skill, rule, or workflow.
When being wrong is expensive
Discovery mode
In regulated or ambiguous work, I map the current reality, name open questions, and establish the human and AI boundary before implementation begins.
When feedback is cheap
Build mode
In prototypes and tools, I build a rough version, test each part, cut scope, and let observed behavior replace extended deliberation.
At a glance
Have a difficult product problem?
Let’s make the system clear.
I work best where AI, complex workflows, and human judgment meet. If that sounds like your product, I would like to hear about it.