Back to Selected Work
Case Study2025
AI AgentsSaaSEvent TechFull-Stack

Grow Together Club

Connecting Global Leaders. Infinite Opportunities.

Role & Services

Product Design, UI/UX Systems, Multi-Agent AI Architecture, Full-Stack Development

Timeline & Year

2025

Deployment Status

Production / Live

Grow Together Club Hero Overview
Grow Together Club — System Overview1920 × 1080 (16:9)
[ At-A-Glance Specification ]

Target Industry

Private member clubs, executive networking, high-stakes gatherings

Core Audience

Founders, investors, C-suite, private club hosts

Stack Foundations

Next.js 16 (App Router), Vercel AI SDK, PostgreSQL

Key Metric

~90 sec

Event Draft Duration

[ Operational Scoping ]

The Problem & The Applied Solution

The Operational Challenge

Fragmented tools, coordination fatigue, and unverified data.

Hosts planning executive dinners and summits juggled 5+ disconnected tools for briefs, venues, menus, schedules, budgets, vendors, marketing, and RSVPs. Planning a single 50-guest dinner consumed hours of manual coordination, while relative date ambiguities and scattered sheets routinely introduced scheduling and budget errors.

Impact: Coordination overhead, high human-error rate, fragmented context.
The Engineering Solution

Autonomous agents, verified state machines, and grounded execution.

A coordinated multi-agent system where a central Orchestrator Agent ingests natural language or voice dictation, enforces calendar-anchored temporal reasoning, and delegates execution across 9 specialized domain agents. Outputs render as interactive generative-UI cards (event previews, budget charts, run-of-show blocks) rather than flat conversational text.

Outcome: Unified operational dashboard, deterministic type safety, automated verification.
[ System Design & Topology ]

Coordinated Multi-Agent Topology

The Orchestrator Agent parses natural language and voice input, resolves relative calendar references against live system time, and coordinates nine specialized domain sub-agents. Destructive and high-impact operations require strict human-in-the-loop authorization before execution.

Multi-Agent System Design

Coordinated Multi-Agent Orchestration Topology

Temporal Anchoring ActiveHuman-in-the-Loop Gates
TXT

Conversational Input

Natural language event briefs & multi-turn dialog

Vercel AI SDK
MIC

Voice-to-Plan Dictation

Hands-free mobile briefing via Web Speech API

Web Speech
Core DispatcherOrchestrator Agent

Interprets natural language intents, anchors relative temporal expressions ("next Friday") to current calendar to eliminate date hallucination, and delegates tasks with typed tool schemas.

toolApproval: "user-approval"

Destructive actions (mass invites, deletion) require host authorization.

[ 9 Specialist Domain Agents : Click to inspect responsibilities & artifact output ]

Active Specialist

Event Agent: Format & Privacy Strategy

Input: Host briefing, event goals, participant criteria

Configures gathering type (In-Person / Virtual / Hybrid), access tier (Public vs Invite-only token), and executive plan synthesis.

Generates UI Artifact:Executive event blueprint, encrypted invite tokens
Render Engine:Generative UI CardsRun-of-Show TimelinesVariance LedgersVIP QR Passes
Postgres + Drizzle ORM (25+ Tables) ↔ Strict Zod Tool Schemas
01

Temporal Calendar Anchoring: Evaluates phrases like 'next Friday' or 'Q3 summit' against real-time system clocks, eliminating past-year or phantom date hallucinations.

02

Human-in-the-Loop Safeguards: Implements toolApproval: 'user-approval' gates for irreversible actions including mass invitation dispatches and event deletion.

03

Generative UI Streaming: Tool call outputs render directly as interactive React cards (budget ledgers, catering matrices, timeline blocks) instead of walls of unstructured text.

04

End-to-End Type Safety: Strict Zod schemas connect Postgres tables, Drizzle ORM queries, Server Actions, and Vercel AI SDK tool parameters.

Visual Language & Brand Tokens

Luxury Private-Club Design System

Headings: Cormorant Garamond (Editorial / Luxury) · Body: Inter (Neutral / High Legibility)

Midnight Obsidian

#070B12

Primary Background

Deep Navy Slate

#0F1B2E

Elevated Surfaces & Cards

Antique Gold

#B8973A

Accent & Status Highlights

Warm Cream

#F9F6F1

Foreground Headings & Contrast

Interaction & Surface Principles:

Frosted-glass surface cards with subtle gold-tinged border highlights
Framer Motion micro-interactions with spring physics
Mobile-first bottom drawers for swift gesture-driven event auditing
Full dark and light theme parity with zero layout shifts
[ Core Functional Capabilities ]

Engineered Capabilities & Workflows

Each capability module represents a discrete, type-safe subsystem designed for resilience and performance.

  • ✦Natural-language event builder translating unstructured conversational prompts into complete event blueprints.
  • ✦Voice-to-plan dictation powered by browser Web Speech API for rapid mobile briefings.
  • ✦Interactive generative UI cards rendering budget graphs, schedules, and guest lists in-stream.
  • ✦Persistent multi-turn conversation threads with contextual memory and guided walkthrough tours.
[ Production Infrastructure ]

Full-Stack Architecture & Technology Stack

Strictly vetted dependencies ensuring type safety, low latency, and zero vendor lock-in.

Layer 01

Frontend & UI

Next.js 16 (App Router)React 19React CompilerTailwind CSSshadcn/uiFramer Motion
Layer 02

AI & Model Pipeline

Vercel AI SDKOpenAI GPT-4oGPT-4o-miniDALL-E 3 (16:9 Art)
Layer 03

Data & Storage

PostgreSQLDrizzle ORM (25+ typed tables)UploadThing
Layer 04

Auth & Security

Better Auth (Sessions)Google OAuthMagic LinksTokenized RSVPs
Layer 05

Infrastructure & Quality

Upstash Redis (Rate Limiting)Resend + React EmailTypeScript Strict ModeBiome
[ Quantifiable Impact ]

Key Engineering Highlights & Outcomes

~90 secInternal estimate

Event Draft Duration

Reduced from ~4 hours across 5 tools to a structured 90-second conversational briefing.

9 Agents

Autonomous Domain Specialists

Orchestrator coordinates 9 specialized sub-agents with dedicated schema validation.

100%

Type-Safe Pipeline

Unbroken type guarantee from Postgres tables and Drizzle ORM to Server Actions and tool calls.

Zero Drift

Temporal Anchoring

Resolves relative scheduling expressions against live calendars, preventing past-year hallucinations.

[ Product Interface Showcase ]

Visual Workspace & Operational Interfaces

Inspect primary screens, interactive cards, and live operational consoles. Click any image to expand full-resolution view.

Grow Together Club Host Operating Dashboard showing event metrics and navigation
Click to Expand Lightbox

Host Operating Dashboard & Event Overview

1920 × 1080 (16:9)

Primary host dashboard displaying the event header, RSVP fill progress, upcoming schedule blocks, and conversational assistant entry point.

Conversational AI event planning interface with live voice dictation indicator
Click to Expand Lightbox

Conversational AI Concierge & Voice Dictation

1600 × 1000 (16:10)

Chat interface demonstrating voice input, multi-turn plan synthesis, and specialist agent task delegation pills.

Interactive generative UI cards showing catering matrix and budget allocations
Click to Expand Lightbox

Generative UI Cards & Budget Ledger

1600 × 1000 (16:10)

Interactive in-stream generative UI cards displaying side-by-side venue quotes, dietary breakdown, and cost variance.

Minute-by-minute run-of-show schedule builder with speaker cue assignments
Click to Expand Lightbox

Run-of-Show Cascading Timeline Builder

1600 × 1000 (16:10)

Detailed agenda block builder showing speaker cue assignments, tech lead notes, and calendar export controls.

VIP mobile digital pass with QR verification code and camera check-in drawer
Click to Expand Lightbox

Digital VIP Guest Pass & Fast QR Scanner

1200 × 1200 (1:1)

Mobile-first attendee pass view featuring antique gold branding, personal QR token, and live check-in validation state.

[ Engineering Partnership ]

Have a similar engineering challenge?

Whether architecting multi-agent generative systems, private on-premise AI, or speech intelligence pipelines — we engineer solutions that perform in production.