AI products / assistants / workflow automation

AI product design that works beyond the prompt

I design AI assistants and workflow products around real actions: what the system knows, what it can do, when a person reviews the result, and how the product recovers when something fails.

What I design

AI assistant UX

Conversation, action, confirmation, recovery, empty, loading, and permission states that make the assistant understandable beyond the first prompt.

Workflow automation

Triage, routing, review, schedules, notifications, and human handoff across the tools where the work already happens.

Product architecture

A coherent product model for sources, actions, system status, user control, and the boundary between automation and human decisions.

Prototype to release

UX/UI, design systems, React prototypes, implementation review, analytics, and release QA in one delivery loop.

Relevant work

AI Assistant / Ops Workflow

Pochtalyon AI

Slack, Google, and YouTrack created a scattered work queue. Missed replies, meetings, and tasks had no one action queue.

2026 private ops product. Scope: architecture, states, source aggregation, and routing.

Read case
Media Product / Match Data + Admin

NextSport.io

Meta-Next had to stop being separate screens and ship as one product: match center, editorial, admin, AI content ops, telemetry, API, mobile shell, and delivery.

2026-01-28 to 2026-06-03. Live at nextsport.io: 139 route nodes, 25-slide proof deck, real UI screens, admin, AI content ops, telemetry, mobile states, and delivery.

Read case

Bring the workflow, not a polished brief

Send the current process, the tools involved, and the point where people lose time or confidence. I can help frame the product, design the states, prototype the interaction, and stay with implementation.

Contact Mikhail