Chat
Chat is the legacy Salesforce web-chat channel, originally called Live Agent, that let customers start a real-time text conversation with a service rep from a website or app.
Definition
Chat is the legacy Salesforce web-chat channel, originally called Live Agent, that let customers start a real-time text conversation with a service rep from a website or app. A visitor clicked a chat button, an Omni-Channel queue routed the request to an available agent, and the whole exchange was saved as a Live Chat Transcript record in Salesforce. It was a core Service Cloud feature for more than a decade.
Salesforce retired Legacy Chat on February 14, 2026. New orgs can no longer set it up, and existing deployments are expected to move to Enhanced Chat, the channel formerly known as Messaging for In-App and Web. If you still see Chat in an older org, treat it as a sunset feature. The sections below explain how classic Chat worked, why it was retired, and how the replacement differs so a migration goes smoothly.
How Chat worked and what replaced it
The four moving parts
Salesforce documentation described web Chat as four pieces working together. The first was the Chat Console, the agent-side workspace inside the Service Console where reps read and sent messages. The second was Omni-Channel, the routing engine that pushed each new chat to a qualified, available agent. The third was Embedded Service, the deployable widget that put a chat button on a public website or Experience Cloud site. The fourth, added later, was Einstein Bots, an automated front end that answered common questions before a human got involved. A working deployment needed at least the console, a routing method, and a button. Bots were optional but common in mature setups. Understanding these four parts matters during migration, because each has a direct equivalent in Enhanced Chat. The console becomes the enhanced conversation window, Omni-Channel routing moves to flow-based routing, the Embedded Service button becomes an Embedded Messaging deployment, and Einstein Bots are replaced by enhanced bots or Agentforce. Mapping old to new component by component is the cleanest way to plan the move.
Chat buttons and Embedded Service deployments
Two configuration objects controlled the customer-facing side. A Chat Button, stored as a LiveChatButton record, defined a single entry point: which queue or skill it fed, the pre-chat form attached to it, and its online and offline behavior. A Deployment held the broader settings for a site, such as branding and which buttons it exposed. Admins generated a short JavaScript snippet from the deployment and pasted it into the website. When a visitor loaded the page, that snippet rendered the chat button and, on click, opened the chat window. Embedded Service was the more modern, Lightning-based way to do this and largely replaced the older raw deployment code. Pre-chat forms hung off the button and collected a name, an email, and an issue category before the conversation began. That data routed the chat to the right team and prepopulated the agent's screen. Skipping pre-chat produced cold starts where the agent had no context, which slowed every resolution.
Omni-Channel routing for chats
Routing decided which agent got which chat. Early Live Agent shipped its own basic routing, but Salesforce later steered all customers toward Omni-Channel, the unified engine that also handles cases, leads, and other work. Omni-Channel matched each chat to an agent by availability, configured capacity, and skills. Capacity is the key control. Each agent has a capacity number, and each chat consumes part of it, so a rep set to a capacity of three handles up to three concurrent chats. Skills-based routing sent specialist or high-tier conversations to the agents trained for them. Without a routing engine, chats fell into a manual queue that agents picked from by hand, which did not scale past a small team. Omni-Channel routing carried forward into Enhanced Chat but in a more flexible form. The replacement uses Omni-Channel flow-based routing, where an admin builds the routing logic in Flow. That allows routing straight to a named rep, branching on conversation data, and handing off cleanly to and from a bot.
Bots as the front line
Most busy Chat deployments did not send every visitor straight to a human. They put a bot in front. Einstein Bots was the automation layer for classic Chat. A bot greeted the visitor, asked what they needed, answered routine questions, looked up order or account data through Apex or flows, and resolved simple issues end to end. Only when the bot could not help did it escalate to a live agent, passing along the transcript so the customer did not repeat themselves. The metric that mattered was containment, the share of conversations the bot finished without human help. A well-tuned bot could contain a large fraction of routine traffic, which freed agents for harder work. The modern path replaces Einstein Bots with enhanced bots or with Agentforce, the reasoning-engine generation of Salesforce automation. Agentforce can interpret a request in natural language, decide which action to take, and complete tasks rather than just match keywords to a script. Migration plans should account for rebuilding bot logic on the new framework.
The Live Chat Transcript record
Every chat left a permanent record. Salesforce stored each session as a Live Chat Transcript, available through the API since version 24.0, with the full message log, the agent who handled it, start and end times, wait time, and any Knowledge articles shared during the conversation. A companion object, the Live Chat Transcript Event, captured discrete moments inside a chat, such as a transfer or the agent accepting the request. Transcripts usually linked to a Case, a Contact, or a Lead, which tied the conversation into the customer's wider history. Because transcripts were standard records, you could report and build dashboards on them like any other object. Common reports covered chat volume by hour, average handle time per agent, customer satisfaction by skill, abandonment rate, and bot containment. Enhanced Chat keeps the same idea but uses a different data model built on Messaging Sessions rather than Live Chat Transcripts. That difference is the single biggest reporting gotcha during migration, because dashboards built on the old object do not automatically cover conversations captured on the new one.
Why it was retired and what changed
Salesforce retired Legacy Chat on February 14, 2026, as part of a broader move to Enhanced Service Channels. The official guidance is to switch to Enhanced Chat, formerly called Messaging for In-App and Web, to avoid a service interruption. The reason is architectural. Classic Chat was synchronous, meaning the conversation lived only while both people stayed connected; close the window and the session ended. Enhanced Chat is asynchronous. A customer can start a conversation, leave, and pick it up later on the same thread, much like texting. The replacement adds estimated wait times, automatic responses when a chat starts or ends or an agent joins, the ability for customers to search their own message history, and push notifications for in-app conversations. Routing moves to flow-based Omni-Channel, and bots move to the enhanced or Agentforce framework. Salesforce also offers a Chat Transition Readiness Report that scans an existing setup, lists active deployments and customizations, and helps gauge how much work the move will take before you start.
Planning a clean migration
Moving off Legacy Chat is a project, not a flip of a switch, and treating it that way avoids surprises. Start by running the Chat Transition Readiness Report to inventory every deployment, button, pre-chat form, and customization in the org. That list is your scope. Next, map each old component to its Enhanced Chat equivalent, since the console, routing, button, and bot all change form even though the customer experience looks similar. Rebuild routing as an Omni-Channel flow rather than copying the old skill rules one to one, because flows can do more and the old structure rarely transfers cleanly. Recreate bots on the enhanced or Agentforce framework and retest every escalation path, because a bot that fails to recognize a request to reach a human creates dead-end conversations. Rebuild reports and dashboards against Messaging Sessions, not Live Chat Transcripts, or your metrics will quietly go blank. Finally, decide how to handle historical transcripts; they remain as records for reporting and audit, but new conversations land on the new object, so plan to report across both during the transition window.
Trust & references
Cross-checked against the following references.
- Legacy Chat Is Being RetiredSalesforce
- Replace Legacy Chat with Enhanced ChatSalesforce
Straight from the source - Salesforce's reference material on Chat.
- Chat with Customers on Your WebsiteSalesforce
- LiveChatTranscript Object ReferenceSalesforce
Hands-on resources to go deeper on Chat.
About the Author
Dipojjal Chakrabarti is a B2C Solution Architect with 29 Salesforce certifications and over 13 years in the Salesforce ecosystem. He runs salesforcedictionary.com to help admins, developers, architects, and cert/interview candidates sharpen their fundamentals. More about Dipojjal.
Test your knowledge
Q1. What is Chat in Salesforce currently branded as?
Q2. How are Chat conversations routed to agents?
Q3. What can be combined with Chat to deflect simple inquiries?
Discussion
Loading discussion…