AI Events Planning Assistant

Stack

• Python
• LLM tool-calling (OpenAI / Anthropic)
• Structured agent loop
• Next.js
• TypeScript
• Tauri
• REST APIs

The Operational Challenge

Event teams juggle venues, vendors, schedules, and guest lists across chats, spreadsheets, and tribal knowledge. Questions that should take seconds — “which venues fit this headcount and date?” or “what’s left for vendor follow-up?” — often meant hunting through tools and threads.

Homeland Iyota’s events division needed a single assistant that could answer in natural language and take structured actions, without forcing planners onto a complex new UI for every workflow.

The Architecture Stack

The system centers on a Python backend that runs an LLM agent loop with explicit tool-calling: the model proposes tools, the server executes them against domain actions (venues, scheduling, vendors, guests), and results are fed back into the conversation in a controlled, auditable loop.

Planners interact through a Next.js web application and a Tauri desktop client that talk to the same backend — so day-to-day ops can run in the browser or as a lightweight desktop app without duplicating business logic.

• Python agent service: structured tool-calling and conversation/state handling
• Next.js front end for web access and operational UI
• Tauri desktop shell for a native-feeling client on the same APIs
• Domain tools for venue suggestions, scheduling, vendor coordination, and guest management

Reliability & Agent Design

Agent systems fail when tools are vague or side effects are uncontrolled. The assistant is built around typed, purpose-built tools and a structured loop so the model can request actions rather than inventing free-form database writes.

Splitting web and desktop thin clients from a single Python brain kept deployment flexible: ops can use the live web login while the desktop path shares the same automation core.

• Tool-calling over free-form replies for operational actions
• Shared backend for web + desktop to avoid divergence
• Production deployment with ongoing maintenance for the events team

The Concrete Outcome

Delivered to production in November 2025 and actively used by the Homeland events team — proving Hassan can ship LLM agent products end-to-end: model orchestration, APIs, Next.js UX, and Tauri desktop packaging.