OpenEvent

AI-Powered Event Management Platform

Automated 80% of manual event coordination, cutting team response time from hours to minutes.

ReactTypeScriptSupabaseOpenAIpgvectorStripeDockerGitHub Actions

The Problem

Event management companies drown in emails, manual booking, and scattered spreadsheets. Staff spend hours triaging requests, copying data between systems, and chasing approvals. The work is repetitive but requires judgment that simple automation can't handle.

The Solution

An AI-powered platform where emails are ingested, classified by intent, turned into tasks, and routed through human-in-the-loop approval workflows. The AI proposes actions; humans review and approve with one click. Approved actions trigger real integrations: calendar bookings, Stripe invoices, CRM updates.

  • Email-to-task pipeline: ingests, deduplicates threads, extracts intent and entities automatically
  • Human-in-the-loop workflows: AI drafts actions, humans approve or edit before execution
  • Declarative workflow engine: JSON/YAML definitions with conditionals, parallel nodes, and LLM steps
  • Semantic search via pgvector for template matching and historical context lookup
  • Full integrations: Google/Outlook Calendar, Stripe invoicing, CRM sync, webhooks

System Architecture

Loading diagram...

Key Technical Decisions

Supabase Edge Functions for AI

LLM calls, entity extraction, and workflow triggers run on Deno-based edge functions. Fast cold starts, built-in auth, and direct database access.

pgvector over external vector DB

Kept vector search inside Postgres to avoid another service dependency. Simpler ops, single source of truth, good enough performance for our scale.

Declarative workflows over code

JSON/YAML workflow definitions instead of hardcoded logic. Non-engineers can modify workflows. Supports conditionals, parallel execution, and human review gates.

Results

80% reduction in manual event coordination tasks
Response time dropped from hours to minutes
Serving active clients in production
Full CI/CD pipeline with staging and production environments