Core Concepts¶

Understand the key concepts behind Functional AI to get the most out of the platform.

The Big Picture¶

Functional AI helps you create AI-powered chatbots that can answer questions, help customers, and perform actions - all without coding. Here's how the main pieces fit together:

[Your Documents] → [Knowledge Base] → [Assistant] → [Chat Widget] → [Your Customers]

Documents: Your product guides, FAQs, policies, etc. Knowledge Base: Stores and searches your documents intelligently Assistant: The AI "brain" that generates responses Chat Widget: What your customers see and interact with

Assistants¶

An Assistant is your AI chatbot configuration. It defines how your chatbot behaves, responds, and helps users.

What an Assistant Contains¶

Name: Display name shown to users (e.g., "Product Support", "Sales Helper")
Instructions: Detailed guidance telling the AI how to behave, respond, and handle different situations
Temperature: Creativity level (lower = more consistent, higher = more varied responses)
Model: Which AI model powers the responses (e.g., GPT-4o, GPT-4o-mini)

Public vs Private¶

Private Assistants: Only you can chat with them in the dashboard. Use private mode for testing and development.
Public Assistants: Can be deployed to customers via widgets, share links, and integrations. Make assistants public when ready to go live.

How Assistants Generate Responses¶

User sends a message
System retrieves relevant information from connected knowledge bases
Assistant uses your instructions + retrieved context + the AI model to generate a response
Response is sent to the user

Knowledge Bases (Stores)¶

A Knowledge Base (also called a Store) is a collection of documents that your assistant can search and reference when answering questions.

Why Use Knowledge Bases?¶

Without a knowledge base, your assistant can only answer based on general knowledge. With a knowledge base:

Answers are based on your specific information
Responses include accurate details from your documents
The AI doesn't make up information - it uses what you've provided

How Knowledge Bases Work¶

Upload your documents (PDFs, Word docs, text files)
Process - System splits documents into chunks and creates searchable embeddings
Connect the knowledge base to your assistant
Query - When users ask questions, the system finds relevant document sections
Generate - The AI uses those sections to craft accurate responses

This process is called RAG (Retrieval-Augmented Generation) - the AI retrieves relevant information before generating its response.

What to Put in Knowledge Bases¶

Good content for knowledge bases:

Product documentation and specifications
Frequently Asked Questions (FAQs)
Company policies (shipping, returns, warranty)
How-to guides and tutorials
Service descriptions and pricing

Templates and Functions¶

Templates are collections of Functions that give your assistant new capabilities beyond conversation.

What Functions Can Do¶

Instead of just chatting, functions let your assistant:

Search your product catalog in real-time
Look up order status
Calculate shipping costs
Query external APIs
Perform business-specific actions

Example: Shopify Template¶

The Shopify template includes functions like:

Function	What It Does
`search_products`	Finds products matching a user's query
`product_details`	Gets detailed info about a specific product
`order_lookup`	Retrieves order status by email and order number
`list_collections`	Shows available product categories

System vs Custom Templates¶

System Templates: Pre-built by Functional AI, ready to use (e.g., Shopify integration)

Custom Templates: Your own collections of functions, either curated from system templates or custom-built

Shared Assistants (Public Chatbots)¶

When you make an assistant public, it becomes a Shared Assistant that can interact with your customers.

Deployment Options¶

Website Widget: Embed a chat bubble on any website using a simple script tag
Share Link: Direct URL that opens your chatbot - share via email, social media, or QR codes
Shopify Integration: Native widget embedded in your Shopify store
Social Media: Connect to Instagram DMs, Facebook Messenger, Gmail

What Users See¶

Customers interact with a chat interface that includes:

Your assistant's name and avatar
Suggested questions (if configured)
Chat history within their session
Typing indicators and message formatting

Messages and Conversations¶

What's a Message?¶

Each AI response counts as one message against your quota. User messages don't count - only responses from your assistant.

What's a Conversation?¶

A conversation is a thread of messages between a user and your assistant. Conversations:

Have a unique identifier
Contain all messages exchanged
Show in your dashboard for monitoring
Include metadata (channel, timestamps, costs)

Conversation Channels¶

Conversations can come from different sources:

Website Widget: Embedded chatbot on your sites
Share Link: Direct assistant URLs
Shopify: Your Shopify store widget
Instagram: DMs and comment replies
Messenger: Facebook Page messages
Gmail: Email conversations

All channels appear in your unified Conversations dashboard.

How Everything Connects¶

Here's how the pieces work together in a typical setup:

[Customer asks: "Do you have blue shoes in size 10?"]
                    ↓
            [Chat Widget receives message]
                    ↓
            [Assistant processes request]
                    ↓
    ┌───────────────┴───────────────┐
    ↓                               ↓
[Knowledge Base]            [Shopify Template]
Search for relevant         Call search_products()
product documentation       for real-time inventory
    ↓                               ↓
    └───────────────┬───────────────┘
                    ↓
        [AI combines all information]
                    ↓
    [Response: "Yes! We have the Nike Air Max
     in blue, size 10. $129.99, 3 in stock..."]

Key Terms Quick Reference¶

Term	Definition
Assistant	Your AI chatbot configuration with instructions and settings
Knowledge Base (Store)	Collection of your documents the AI can search
Template	Collection of functions for additional capabilities
Function	Individual action an assistant can perform
RAG	Retrieval-Augmented Generation - searching docs before responding
Shared Assistant	Public-facing chatbot deployed to customers
Conversation	Complete thread of messages with a user
Message	Single AI response (counts against quota)
Widget	Chat interface embedded on a website
Channel	Source of conversations (widget, Shopify, Instagram, etc.)

Next Steps¶

Now that you understand the concepts: