Live chat is real-time, text-based messaging between a business and the people on its website, app, or product. A customer opens a chat window (usually a launcher in the corner of the page), types a question, and gets an answer in the same conversation, without picking up the phone or waiting on an email thread.
The reply can come from a human agent, a rule-based chatbot, or an AI agent. The strongest setups combine AI and people: an AI agent resolves what it can as soon as the conversation starts, then hands anything it can't close to a person in the same helpdesk, with the full history attached.
Key takeaways:
- Live chat is a channel, not a type of software. The chat window is only the surface. Whoever (or whatever) answers behind it decides the experience.
- It runs in five steps: the widget loads, the customer writes in, the conversation is routed, it's resolved or handed off, and the record is saved for reporting.
- You can staff it three ways: human-only, chatbot-first, or AI agent + human. Each has its own cost profile and its own way of failing.
- Pricing usually combines per-seat fees for human agents with usage fees for automation (for example, per AI resolution or outcome).
- The biggest operational risk is the handoff. When your chat tool and your helpdesk are separate tools, context gets lost every time a conversation moves from bot to human.
What is live chat?
Live chat is a messaging channel that lets customers talk to a business in real time from wherever they already are: a product page, checkout, a pricing page, a web app, or a mobile app.
Three things separate it from other channels:
- It's synchronous (or close to it). The customer expects a reply while they're still on the page, not hours later.
- It's in context. The conversation happens on the page the customer is looking at, so the business can see where they are and what they were doing.
- It's persistent. Unlike a phone call, it leaves a written record that the customer, the agent, and any AI on the conversation can refer back to.
What does "live" mean in live chat?
"Live" means real time: messages are sent and answered while both sides are present. It used to mean a human was on the other end. Now it means a live response, whether it comes from a person, an automated flow, or an AI agent, as long as the customer doesn't have to leave and come back later.
Scope: "live chat" also describes consumer chat rooms, streaming chat, and messaging apps. This guide covers live chat as a business customer service and sales channel.
How does live chat work?
Every live chat setup, from a free widget to an enterprise helpdesk, runs through the same five steps.
| Step | What happens | What to get right |
|---|---|---|
| 1. The widget loads | A small JavaScript snippet adds a chat launcher to your site or app. | Load it asynchronously so it doesn't slow the page. Identify logged-in users so conversations carry account context. |
| 2. The customer starts a conversation | They type a question in natural language, or you start the conversation proactively based on behavior (time on page, URL, cart value). | Set expectations upfront: who's answering and how fast. |
| 3. The conversation is routed | Rules decide who takes it first: a bot, an AI agent, or a team inbox. | Route by intent, language, plan, or page, not just "next available agent." |
| 4. It's resolved or handed off | The first responder answers, takes an action, or escalates to a human. | The handoff should carry the full conversation and customer data so nobody repeats themselves. |
| 5. It's logged and measured | The transcript, tags, and outcome are saved to the helpdesk for reporting and quality review. | Measure resolution, not just response time. |
Our live chat software page simplifies this flow into three steps:
- The customer opens Messenger and asks a question in natural language
- Fin AI Agent "responds conversationally using your approved knowledge, citing sources and guiding the customer to resolution"
- When a customer needs human support, Fin hands off with full context, so agent never restarts the interaction.
Reactive vs. proactive live chat
- Reactive chat waits for the customer to open the widget.
- It's the default for support.
- Proactive chat starts the conversation for them: a message on a pricing page, an offer at checkout, a tip inside a new feature.
- It's most useful for sales and onboarding, and most annoying when it fires on every page. Trigger it on intent signals, not timers.
Who answers a live chat? Humans, chatbots, and AI agents
A live chat is answered by a human agent, a rule-based chatbot, or an AI agent working alongside humans. That choice drives your cost and coverage more than any widget feature.
| Human-only live chat | Rule-based chatbot + human | AI agent + human (one helpdesk) | |
|---|---|---|---|
| Who answers first | A support or sales agent | A scripted bot with buttons and decision trees | An AI agent that understands free-form questions |
| Coverage | Business hours, limited by headcount | Business hours, limited by headcount | 24/7 for anything covered by your knowledge and workflows |
| Handles complex questions | Yes | Rarely; it escalates or loops | Often, including multi-step actions; escalates the rest |
| What the customer feels | Personal, but they may wait in a queue | Fast, but rigid; "talk to a human" is a common exit | Fast and conversational, with a human available |
| Handoff quality | n/a | Depends on how the bot connects to the helpdesk | Best when the AI and humans share one inbox and record |
| Cost model | Per agent seat | Seats + bot platform fee | Seats + per resolution/outcome |
| Main risk | Queues and missed chats at peak times | Deflection that frustrates customers | AI answering beyond approved knowledge, so guardrails matter |
Our chatbot vs live chat comparison goes deeper on speed, cost, and customer experience, including the AI agent option most comparisons leave out.
What is live chat used for?
Businesses use live chat for support, ecommerce, in-product help, sales, and onboarding. In short: anywhere a question stands between a customer and their next step.
- Customer support. How-to questions, troubleshooting, account and billing issues: anything a customer would otherwise email about.
- Ecommerce. Order status, returns, shipping, sizing, and checkout problems, often while the shopper still has items in their cart. (See Intercom for ecommerce customer support.)
- In-product support for SaaS. Help inside the app, so users don't leave the workflow they're stuck in. (See Intercom for SaaS customer support.)
- Sales and lead qualification. Buyer questions on pricing and product pages, qualifying interest, and routing high-intent conversations to the right rep.
- Onboarding. Guiding new customers through setup with proactive messages and quick answers.
The Intercom live chat software page groups these into four "support motions": customer support teams, SaaS companies, ecommerce teams, and sales and growth teams. For ecommerce, it points to Fin for order status, returns, product questions, shipping, and checkout issues.
Benefits of live chat (and the trade-offs)
Live chat's main benefits are speed, context, concurrency, conversion, a written record, and scale with AI. Each one comes with an operational cost to plan for.
| Benefit | Why it matters | The trade-off to plan for |
|---|---|---|
| Speed | Customers get help while they're still on the page. | Speed expectations are high; a slow human queue in a "live" channel feels worse than a slow email. |
| Context | You can see the page, account, and history behind the question. | Only if the widget is connected to your helpdesk and customer data. |
| Concurrency | One agent can handle several chats at once, unlike phone. | Too many concurrent chats hurts quality. Set limits. |
| Conversion | Answering buyer questions in the moment removes blockers to purchase. | Proactive messages can feel pushy if they're poorly targeted. |
| A written record | Transcripts feed quality reviews, training, and AI knowledge. | You need a policy for data retention and sensitive information. |
| Scalability with AI | An AI agent can resolve repetitive questions around the clock. | It needs good source content and clear handoff rules to stay accurate. |
What are the disadvantages of live chat?
The main ones:
- Customers expect near-instant replies.
- Coverage is hard to staff across time zones.
- A poorly configured bot can trap customers in loops.
- A widget that isn't connected to your helpdesk creates a second inbox your team has to watch.
Most of these are design problems rather than channel problems. Good routing fixes some. An AI agent up front fixes coverage. Keeping humans and AI in one helpdesk fixes the rest.
Live chat vs. other support channels
Live chat beats email on speed and phone on cost per contact, but each channel still has jobs it does better.
| Channel | Best for | Response expectation | Where live chat beats it | Where it beats live chat |
|---|---|---|---|---|
| Email / tickets | Detailed or asynchronous issues, attachments, follow-ups | Hours | Speed and in-context help | Long, multi-day investigations |
| Phone | Urgent, emotional, or high-stakes issues | Immediate | Concurrency, written record, lower cost per contact | Complex conversations where tone matters |
| Help center | Self-serve answers to common questions | Instant (if the article exists) | Handles follow-ups and edge cases | Zero contact cost when the article answers it |
| Social and messaging apps (WhatsApp, Instagram, etc.) | Customers who prefer their own apps | Minutes to hours | In-product and on-site context | Reaching customers off your site |
You don't have to pick one. Intercom routes "email, live chat, phone, SMS, WhatsApp, Instagram and Facebook" to the same Inbox, so live chat is one channel in a single queue, not a separate tool. Here's how omnichannel customer service works.
What to look for in live chat software
Judge live chat software on what happens behind the widget, not the widget itself. Six questions cover it:
- Who answers first. Is there an AI agent that can resolve questions from your knowledge, or only a rule-based bot?
- The handoff. When a human takes over, do they see the full conversation, what the AI already tried, and the customer's account data?
- One inbox. Do chat, email, and other channels land in the same queue, or is chat a separate tool?
- Brand control. Can you control the widget's design, tone, and behavior on web and mobile?
- Reporting. Can you see AI and human performance side by side: resolution rate, CSAT, and reopen rate?
- Pricing model. Per seat, per conversation, per resolution, or a mix? And how does the bill change as volume grows?
Our guide to the best live chat software scores the main tools against these criteria. Already on a chat tool and thinking of switching? Start with live chat alternatives.
How to add live chat to your website
For most platforms, adding live chat takes three steps: create an account, customize the widget, and paste a JavaScript snippet into your site (or install a plugin for WordPress, Shopify, or Google Tag Manager). The harder decisions come after: who answers, how you identify logged-in users, and when a conversation goes to a human.
Our step-by-step guide on how to embed a live chat widget on your website covers each install method, a worked example with the Intercom Messenger, and a pre-launch testing checklist.
How much does live chat cost?
Live chat is priced in four main ways, and many vendors combine them:
- Free plans with limits on features, history, seats, or branding.
- Per-seat pricing, where every agent who answers chats needs a paid license.
- Bundled or per-workspace pricing, where a flat monthly fee includes a set number of seats or conversations.
- Usage-based AI pricing, charged per AI resolution, conversation, or outcome on top of seats.
Intercom combines seats and outcomes. The pricing page lists plans from $29 per seat per month (Essential, with a new-customer price of $19 shown at the time of writing), and Fin AI Agent on every plan from $0.99 per outcome. An outcome is counted when Fin resolves the customer's issue or completes a configured workflow, including a handoff, and you are charged at most once per conversation, even when Fin answers several questions. Fin does not charge for unsuccessful attempts.
Our live chat pricing guide puts published prices side by side, including what each vendor counts as an "AI resolution."
How Intercom approaches live chat: Messenger, Fin, and one Inbox
Intercom runs live chat through Messenger, with Fin AI Agent answering first and your team picking up handoffs in the same Inbox. From the live chat page: Fin "resolves conversations the moment they start, answers free-form questions using your knowledge, and hands off to a human agent with full context when one is needed. Live chat becomes a resolution channel instead of another queue for your support team to manage."
Intercom is the only helpdesk with a natively integrated AI Agent, so AI and human agents work from one record.
The three pieces:
- Messenger is the chat surface on web, iOS, and Android, with control over tone, style, and design.
- Fin is the AI agent that answers first. On Messenger, Fin replies without quick-reply buttons, follows up if a customer goes quiet, and shows sources in its replies.
- Fin resolves 76% of queries on average, including complex, multi-step work (Fin, 2026). More on Fin AI Agent and the Fin Live Chat product page.
- Intercom is where humans pick up. From the Intercom page: "Fin and your team work from the same Inbox, with the same customer history and the same knowledge base." When Fin needs a decision, it can pause, bring in a teammate for a review, some information, or an approval, and then finish the job (Human in the Loop).
Contrast that with a bolt-on setup: a standalone chat widget is a separate tool attached to your helpdesk, so context gets lost every time a conversation hands off. The handoff is where most live chat experiences break. Test it in any tool you evaluate.
See Intercom live chat software →
Frequently asked questions
What is live chat in simple terms?
Live chat is a chat window on a website or app that lets customers message a business and get answers in real time. The reply can come from a human agent, a chatbot, or an AI agent.
How does live chat work?
A JavaScript snippet adds a chat widget to your site. When a customer starts a conversation, routing rules send it to an AI agent, a bot, or a team inbox. The first responder resolves it or hands it off, and the conversation is saved in your helpdesk for reporting.
What does "live" mean in a chat?
It means the conversation happens in real time, while the customer is still present. "Live" used to imply a human on the other end; today it describes the speed of the response, which can come from a person or an AI agent.
Is live chat with a real person?
It can be. Some businesses staff live chat entirely with people; others put a chatbot or AI agent first and hand off to a person when needed. A good setup tells customers who they're talking to and makes it easy to reach a human.
Is live chat the same as a chatbot?
No. Live chat is the channel; a chatbot is one way to automate replies in that channel. AI agents are a newer option that understand free-form questions and can resolve issues, not just route them. See chatbot vs live chat for a full comparison.
How much does live chat cost?
It ranges from free plans with limits to per-seat plans for each agent, plus usage-based fees for AI. Intercom, for example, lists plans from $29 per seat per month (Essential; a $19 new-customer price is shown as of October 2026) and Fin AI Agent from $0.99 per outcome. See our live chat pricing guide for more.
What is a live chat widget?
It's the chat launcher and message window that appears on your website or app. It's installed with a code snippet or plugin and connects to the software your team (and any AI agent) uses to reply. Here's how to embed one.
Do I need a helpdesk to use live chat?
Not to get started, but it matters more as you grow. Without a helpdesk, chat becomes a separate inbox and handoffs lose context. With chat inside the helpdesk, every channel and every handoff lives in one place.