Meet Aedan: Your AI Conversation Expert

Not a chatbot. Not a script. Aedan is powered by 5 specialized AI agents working as one intelligent team to have real conversations that convert.

Chat With Aedan Now See Aedan's Results

Hi! I'm Aedan, and I love talking to your customers.

Friendly but professional

Never forgets a customer

Learns your restaurant's unique voice

Available 24/7, never has a bad day

Handles 100+ conversations simultaneously

Why Multi-Agent Architecture Matters

5 Agents, 1 Intelligent System

Orchestrator, Reasoning, Flow Manager, Data Extractor, and Validator agents collaborate in real-time. Each agent has a specific job, ensuring accurate, reliable responses every time.

Persistent Conversation Memory

Every conversation is tracked in Firebase with session IDs. Your AI remembers customer preferences, order history, and context across multiple interactions.

Real-Time WebSocket Messaging

Powered by Socket.io gateways for instant, bidirectional communication. No polling delays - messages appear instantly with sub-100ms latency.

Multiple Gemini Models

Access to 5 Google Gemini models based on your plan: 2.0-flash-lite, 2.0-flash, 2.5-flash-lite, 2.5-flash, and 2.5-pro. Higher tiers unlock more powerful reasoning.

Scenario 1: Complex Orders

Customer: "What's your spiciest pasta with no dairy?"

Aedan's Thought Process:

  1. Reasoning Agent identifies: spicy + pasta + dairy-free
  2. Data Extractor scans menu for matches
  3. Validator checks ingredient lists
  4. Flow Manager prepares upsell options

Aedan: "Our Arrabbiata is our spiciest pasta and naturally dairy-free! It's made with crushed red pepper, garlic, and San Marzano tomatoes. Want to add grilled chicken for $5?"

Result: Order placed, dietary needs met, $5 upsell

Scenario 2: Customers Change Their Mind

Customer: "Actually, can I change my reservation from 7pm to 8pm?"

How Aedan Handles It:

  • Checks Chrissy (Reservations Manager) for 8pm slots
  • Finds table for 4 available at 8:15pm
  • Offers alternative: "8:15pm works perfectly"
  • Updates reservation automatically
  • Sends new confirmation email

Result: Seamless modification, customer happy

Scenario 3: Rush Hour Performance

Friday, 7:00 PM - Dinner Rush Begins

Aedan's Dashboard:

47
simultaneous conversations
12
orders being placed
8
reservations being modified
0
customers waiting

Average Response Time: 1.2 seconds

Conversion Rate: 78%

Revenue During Rush: $3,247

Result: Zero missed orders, perfect service

Aedan's "5-Agent Brain"

Every claim below is backed by actual code from our production system. No marketing fluff.

Agentic Flow Orchestrator

/backend/src/agents/agentic-flow-orchestrator.service.ts

The conductor of the AI symphony. This agent coordinates all other agents, manages conversation flow, and decides which specialized agent should handle each part of the customer interaction. It maintains conversation state and ensures coherent, multi-turn dialogues.

Reasoning Agent

/backend/src/services/gemini.service.ts (structured reasoning mode)

Uses advanced semantic understanding to interpret customer intent. This agent doesn't just pattern-match keywords - it actually understands context, disambiguates unclear requests, and reasons about what the customer truly wants.

Flow Manager

/backend/src/agents/agentic-flow-orchestrator.service.ts

Manages conversation paths and decision trees. Determines if a customer is ready to book a reservation, needs more menu information, or wants to modify an existing order. Prevents conversation dead-ends and ensures smooth transitions.

Data Extractor Agent

/backend/src/services/ai-data-extraction.service.ts

Extracts structured data from natural language. Pulls out customer names, phone numbers, reservation times, party sizes, menu items, and special requests from conversational text with high accuracy confidence scoring.

Validator Agent

/backend/src/controllers/agent.controller.ts (lines 469-497)

Validates extracted data against restaurant constraints. Checks menu item availability, verifies reservation time slots, validates party sizes against table capacity, and ensures all data meets business rules before booking.

How It Works: Step-by-Step

1

Customer Initiates Conversation

A customer messages your AI widget. The system generates a unique session ID and creates a conversation thread in Firebase. All messages are stored with timestamps for persistent context.

2

Orchestrator Analyzes Intent

The Flow Orchestrator receives the message and analyzes conversation history. It determines customer intent (reservation, order, question) and routes the request to the appropriate specialized agents.

3

Reasoning Agent Processes Context

The Reasoning Agent uses Google Gemini models (2.0-flash through 2.5-pro depending on your plan) to understand semantic meaning, disambiguate unclear requests, and formulate an intelligent response strategy.

4

Data Extraction & Validation

If booking data is detected, the Data Extractor pulls structured information (names, times, items). The Validator checks against restaurant constraints (table availability, menu inventory, hours).

5

Real-Time Response via WebSocket

The final response is sent through WebSocket gateway for instant delivery. Customer sees the reply in real-time with no polling delays. Conversation state is updated in Firebase for future context.

Conversation Limits by Plan

All plans include the full 5-agent system. Higher tiers unlock more powerful Gemini models and higher limits.

Free Trial

25

conversations/month

  • Gemini 2.0-flash-lite
  • 5-agent system
  • WebSocket real-time
  • Conversation memory

Starter

500

conversations/month

  • Gemini 2.0-flash
  • Everything in Free
  • Priority processing

Professional

1,000

conversations/month

  • Gemini 2.5-flash
  • Gemini 2.5-flash-lite
  • Advanced reasoning
  • Everything in Starter

Growth

5,000

conversations/month

  • Gemini 2.5-pro
  • Premium reasoning
  • Everything in Professional

Enterprise

20,000

conversations/month

  • All Gemini models
  • Dedicated support
  • Custom integrations

Frequently Asked Questions

How is this different from a chatbot?
Traditional chatbots use keyword matching and rigid decision trees. AEDAN uses 5 specialized AI agents with semantic reasoning powered by Google Gemini. It actually understands context, maintains conversation state, and adapts to customer needs in real-time.
Which Gemini models do I get access to?
Free tier: Gemini 2.0-flash-lite. Starter: 2.0-flash. Professional: 2.5-flash and 2.5-flash-lite. Growth and above: Gemini 2.5-pro. Higher-tier models have better reasoning, context understanding, and accuracy.
Does conversation memory persist across sessions?
Yes. Every conversation is stored in Firebase with a unique session ID. When a customer returns, the AI can access previous conversation history, preferences, and order data. Session IDs are tracked in `/backend/src/controllers/agent.controller.ts` (lines 184-210).
What happens when I hit my conversation limit?
The AI displays a friendly upgrade message to customers. Restaurant owners can log into the admin dashboard to upgrade their plan instantly. No service interruption - just a clear upgrade path. Limits reset on the 1st of each month.
Can I customize the AI's personality or responses?
Yes. In your restaurant settings, you can set a custom prompt that defines the AI's tone, personality, and behavior. The Flow Orchestrator uses this prompt as part of its reasoning context. Found in `/backend/src/controllers/restaurant.controller.ts`.
How fast are responses?
Real-time via WebSocket with sub-100ms latency once the AI processes the request. Gemini 2.0-flash-lite responses average 1-2 seconds. Gemini 2.5-pro (Growth tier+) averages 2-4 seconds due to more sophisticated reasoning.
What data does the AI have access to?
The AI accesses your restaurant profile (name, cuisine, hours), menu data (from PDF training), reservation availability, order history, and conversation context. All data is restaurant-specific and isolated - no cross-contamination between restaurants.
Is there an API for custom integrations?
Yes. Growth and Enterprise tiers include API access. You get public and private API keys (found in `/backend/src/controllers/restaurant.controller.ts` lines 948-987) for integrating AEDAN into your own apps or systems.

Ready to Experience Real AI Conversations?

Start with 25 free conversations. No credit card required.

Try AEDAN Now