liildev

Work detail

Event planning & marketplace

From an event idea to a plan, providers, and a shared workspace — with AI assistance throughout.

Listo brings event planning and service discovery into one product. I designed and built it end-to-end: the user experience, full-stack application, structured AI assistants, and production infrastructure.

Role
Product Designer & Engineer
Year
2026
Status
active
Website
listoo.uz →

01

Problem

Planning an event means coordinating providers, schedules, guests, and costs across separate tools. Listo connects the journey: describe an idea, clarify the plan with AI, discover relevant services, then manage everything in a dedicated event workspace.

02

What I built

  1. 1.AI Planner: conversational setup that maintains event type, city, date, guest count, budget, and required services. It clarifies missing details and recommends relevant providers as the plan develops.
  2. 2.Event workspace: an AI assistant alongside providers, programme, tasks, budget, guests, and event readiness. The assistant works with the current event and respects confirmation boundaries for changes.
  3. 3.Provider marketplace: city-based discovery, categories, and filters for venues, restaurants, photographers, transport, entertainment, decorators, and more. Profiles bring together media, pricing units, reviews, saved providers, and recommendations, with availability and booking where supported.
  4. 4.Partner area: business profiles, services and packages, pricing, media, locations, served cities, leads, booking requests, and organization/team management.
  5. 5.Event website: a publishable invitation with programme and guest-facing event details, connected to the planning workspace.

03

Architecture

  1. 1.Full-stack TypeScript: React 19 with TanStack Start, Router, and Query; Tailwind CSS and a shared UI system; Hono and tRPC for the API; PostgreSQL 16 with Drizzle ORM for product data.
  2. 2.Structured AI turns: Planner and Event assistants share an architecture that separates structured understanding from the final user-facing response. Both use GPT-6 Luna; PostgreSQL remains the canonical product state.
  3. 3.Application infrastructure: Bun, Redis, and Inngest support the runtime and background workflows. Cloudflare R2 stores media.
  4. 4.Production delivery: Docker, CircleCI, and Cloudflare Tunnel, with automated quality/release checks, expand → rollout → contract migrations, backups, and restore verification.

04

UX decisions

  1. 1.Start with a conversation instead of a long form. Clarify missing details progressively while preserving the structured event context.
  2. 2.Keep discovery connected to the plan: city, services, and event context help users find relevant providers without restarting their search.
  3. 3.Give each event one workspace for planning and coordination. Keep the guest-facing invitation focused on information guests need.
  4. 4.Keep users in control: Event AI assistance respects confirmation and authorization boundaries when changing event data.

05

Challenges

  1. 1.AI state continuity: preserve the event plan across conversations and recommendations while keeping database state authoritative rather than relying on model conversation history.
  2. 2.Two connected product sides: support planning and provider discovery for customers alongside business, service, lead, and booking management for partners.
  3. 3.Production ownership: carry product architecture and UX/UI through frontend, backend/API, database design, migrations, AI integration, testing, CI/CD, and release engineering.

06

Outcome

Live at listoo.uz. Listo brings conversational planning, provider discovery, event coordination, partner tools, and guest invitations into one production product. My role spans the complete system, from product design to deployment and ongoing engineering.

Discuss a project→