Artificial Intelligence is now a core part of modern customer support. In Crisp, AI is not an add-on, but a complete system built to reduce repetitive work, improve response quality, and help teams deliver faster, clearer support at scale.

This guide will help you understand how Crisp AI works, how each layer contributes to your workflow, and how to start using it effectively from day one.

Whether you're an admin or an operator, this article walks you through the essential concepts in a simple, structured way.

The AI Data Hub: Foundation of Every AI Feature

Before using any AI function in Crisp, it’s important to understand the *AI Data Hub*, which is the central engine behind all automated answers.

Instead of relying on external models, Crisp AI learns directly from your own sources of truth:

• your website content,
• your knowledge base,
• your previous conversations,
• your imported documents (.csv, .pdf, .txt),
• and your internal snippets.

The Data Hub continuously analyses and restructures this information so the AI can provide accurate, contextual answers across every feature in the product. Because the model is trained on your own content, it remains consistent with your tone, your processes, and your policies.

To get the best results, ensure your website and knowledge base are up to date, and enrich your internal snippets whenever you see information your team often needs to repeat. The more complete your Data Hub becomes, the more reliable the AI will be across all layers.

How can I add data sources to Crisp AI Hub?

Layer 1: Reduce Incoming Questions With the Chatbox Search AI

The Chatbox Search AI is the first layer of your AI stack. It appears inside your Crisp Chatbox and lets visitors search for answers *before* starting a conversation.

Instead of browsing articles manually, users can ask questions in natural language and receive instant answers generated from your AI Data Hub. This helps filter out repetitive questions and reduces the number of conversations entering your Inbox.

How to enable the Chatbox Search AI

To activate the Chatbox Search AI:

1. Go to *Settings → Chatbox Settings → Chatbox Search AI*
2. Toggle *Enable Crisp Chatbox Search AI on my Crisp Chatbox widget*
3. The Chatbox Search AI becomes immediately visible inside your chatbox

How to configure it properly

To get the best results from the Chatbox Search AI:

• Add *Frequent Searches* so users can click on common questions directly
• Add *Categories* to help users navigate your support topics if you cover multiple products or areas
• Keep your *Data Hub* clean and updated, since all answers come directly from the data you index

How to set up Crisp Chatbox Search AI?

Recommendation

For most teams, the Chatbox Search AI is the easiest and most impactful AI feature to enable first. Once configured, it can deflect 30-50% of simple requests and give visitors immediate answers before contacting your team.

Strategies to reduce time to resolution with examples from companies around the world

How to offer 24/7 support without outsourcing?

Layer 2: Automate Answers With the AI Agent

Once your AI Overlay is filtering incoming questions, the next layer of your setup is the AI Workflows for customer service automation.

This bot uses the information stored in your Data Hub to answer questions automatically, guide users through steps, or route conversations to the right team. It works on all the channels you’ve connected to Crisp, from your website chatbox to social messaging apps.

To create your workflow, open *AI Agent → Automate → Workflow builder*

This scenario gives you a pre-built structure where the bot replies autonomously, waits for user feedback, and escalates to your team if needed.

Inside the builder, you’ll see that the flow is very simple: a user message triggers MagicReply, the AI generates an answer, and the conversation continues automatically.

You can extend the scenario by adding buttons, fallback messages, escalations, or conditions if you want more control.

How the AI Chatbot really works?

11 AI chatbot best practices for always-on customer service

Ticket deflection: How AI Chatbots reduce support backlogs

Layer 3: Understand Conversations Faster With Summaries and Topics

Once a conversation reaches your inbox, Crisp AI helps you get up to speed instantly.

The *AI Summary* gives you a clear, one-paragraph recap of the entire exchange, so you don’t need to scroll through long threads to understand what’s going on.

Crisp also identifies *key topics* from the conversation and displays them in the sidebar. This lets you see the customer’s intent at a glance and quickly spot recurring issues over time.

These insights reduce cognitive load and make it easier for operators to jump straight into the conversation with full context.

Layer 4: Reply Faster With MagicReply

MagicReply uses your *AI Data Hub* to generate instant reply suggestions directly inside the message composer.

Because it relies only on your own content (help articles, website pages, snippets, documents, past conversations), every suggestion stays accurate and aligned with your support guidelines.

MagicReply can help you:

• Draft clear, ready-to-send replies based on the user’s question
• Maintain consistent answers across your whole team
• Respond faster to factual or repetitive requests
• Reduce cognitive load by giving you a structured starting point
• Adapt the tone of the suggestion to your needs (friendlier, more formal, etc.)

Any troubles with MagicReply?

Layer 5: Write Better Answers With Copilot

Crisp Copilot is your internal AI assistant, designed to help operators answer faster and with greater confidence.

Unlike MagicReply, which suggests direct responses inside the composer, Copilot lives in a dedicated sidebar, meaning it never writes to the customer, only for the agent.

Copilot can assist you all throughout a conversation by clarifying complex topics using your documentation and answering internal questions about your product or processes. It searches across all your public and private data sources to provide accurate, context-aware explanations, and can optionally use the ongoing conversation to tailor its suggestions.

Because Copilot interacts exclusively with the agent, not the customer, it’s a safe and powerful tool for teams that want AI assistance without automating replies.

How to use Copilot in your Inbox

1. *Open any conversation →* Click Copilot in the right sidebar.
2. *Ask your question* → Type anything: a user query, a feature question, or ask Copilot to draft a reply.
3. *Enable conversation context (optional)* *→* When activated, Copilot uses the messages already exchanged in the conversation to tailor its suggestions.
4. *Copy the answer into your reply* → You stay in control. Copilot never sends messages directly to the customer.

Copilot enhances your team’s expertise instead of replacing it, giving every operator instant access to your company’s knowledge, right inside the Inbox.

Conclusion

Crisp AI is built as a complete system, designed to support your team at every stage of the customer journey.

The Data Hub provides the foundation, the Overlay reduces simple requests, the AI Chatbot automates the first line of support, Summaries and Topics help operators understand conversations quickly, MagicReply accelerates replies, and Copilot helps refine or generate messages effortlessly.

By activating these layers progressively and maintaining high-quality data, you can build a support experience that is faster, more reliable, and easier to scale, without losing the human touch that makes great customer service unique.

Check out these helpful resources:

Nail 80% AI-resolved conversations through human-grade support with AI

Overnight support for SaaS companies: AI as the night shift

Customer Service Automations Every Support Team Should Deploy

Bijgewerkt op: 12/02/2026