Salesforce's newest Agentforce release lets agents return Lightning Web Components — not just text — inside the conversational panel. Branded cards, pickers, signature pads and KPI widgets, all natively rendered, all typed, all reusable.
Until now, Agentforce replies were limited to plain text and markdown. Anything richer — booking a slot, signing a quote, picking an asset — meant pushing the user out of the panel into a record page or modal.
Custom Lightning Types (CLT) let an Agentforce Action return a typed payload that is rendered by a Lightning Web Component inside the conversational panel itself. The agent stays in conversation, the user stays in flow, and the team reuses the same component across Service, Sales, Mobile and Embedded channels — with the Trust Layer, permissions and masking automatically applied.
Two illustrative views — a typed BookingPicker rendered inside the conversational panel, and the underlying CLT definition + LWC renderer.
Figure 1 — A typed CLT (BookingPicker) rendered inside the Agentforce panel. Selecting a slot is a new conversational turn (illustrative).
// 1. Define the typed payload (CLT metadata)
@CustomLightningType('BookingPicker')
export class BookingPickerType {
@field({ required: true }) caseId: string;
@field({ type: 'list' }) slots: Array<{ start: string; engineerId: string }>;
}
// 2. Build the LWC renderer
import { LightningElement, api } from 'lwc';
export default class BookingPicker extends LightningElement {
@api payload; // typed as BookingPickerType
@api respond; // sends a new turn back to the agent
pick(e) {
this.respond({ chosenSlot: e.target.dataset.start });
}
}
// 3. Bind it to your Agentforce Action's response type
// → Agent renders it natively in the conversational panel.Figure 2 — Defining the CLT schema and binding the LWC renderer (illustrative).
CLT closes the last-mile UX gap between "the agent told me what to do" and "the agent did it with me, right here in the panel."
The shortest path from a use case to a typed component rendered inside the Agentforce panel.
// 1. Define the typed payload (CLT metadata)
@CustomLightningType('BookingPicker')
export class BookingPickerType {
@field({ required: true }) caseId: string;
@field({ type: 'list' }) slots: Array<{ start: string; engineerId: string }>;
}
// 2. Build the LWC renderer
import { LightningElement, api } from 'lwc';
export default class BookingPicker extends LightningElement {
@api payload; // typed as BookingPickerType
@api respond; // sends a new turn back to the agent
pick(e) {
this.respond({ chosenSlot: e.target.dataset.start });
}
}
// 3. Bind it to your Agentforce Action's response type
// → Agent renders it natively in the conversational panel.Every Agentforce release this year has strengthened a different layer of the same stack. CLT closes the loop on the user-facing surface.
Salesforce is positioning Agentforce as the default agentic surface on top of the Customer 360. Builder made agent assembly low-code; MCP made tool-use open; Data Cloud made grounding native; CLT now makes the user experience as rich and reusable as a Lightning record page. Together they push every layer of the stack — orchestration, reasoning, grounding, and now UI — into the same governed platform, so customers can scale from one agent to many without building a parallel UX stack on the side. For CIOs evaluating "build vs. buy" agent platforms, CLT removes one of the last reasons to wrap Agentforce inside a custom front-end.
Hand-picked Salesforce learning content to get your team CLT-ready.
Official developer documentation covering Actions, MCP tools, Data Cloud grounding and the broader Agentforce extensibility model.
Salesforce's official overview of the Agent Builder where CLT-typed Actions are wired up.
Foundational module on Topics, Actions and the canvas — essential before adding CLTs.
Hands-on quick-start that walks you through building, grounding and deploying a working Service agent.
How CLT plugs into KVP's broader Agentforce playbooks and live engagements.
Salesforce's visual agent studio explained — Topics, Actions, MCP and one-click deploy.
Why rule-based bots stall and how autonomous agents change the math.
A practical developer's playbook for Agent Script, Actions and MCP tools.
Connect Salesforce with Claude using MCP, Agentforce and Headless 360 — 5 use cases.
How a $80B+ growth-equity firm deployed Agentforce across deal flow and IR.
A unified Salesforce platform powering an investor-portal experience.
A 6-slide, social-ready summary. Download as PDF for LinkedIn, or share a single slide.

Rich, interactive UI — right inside the conversation.
Custom Lightning Types (CLT) let agents return Lightning Web Components instead of plain text — quote cards, signature pads, asset finders, KPI widgets — rendered natively in the Agentforce conversational panel.




The same pattern as Builder + MCP + Data Cloud — open, typed, reusable.

KVP brings the Topics, the LWC pattern library, the typed payload schemas and the governance. You bring the use case.

A single-image summary built for LinkedIn, X and WhatsApp. Download or share in one tap.

Render rich, interactive Lightning Web Components inside the Agentforce conversational panel — typed payloads, branded UI, structured user input back to the agent.
Talk to KVP's Agentforce pod about a focused 6–8 week sprint — typed CLT pattern library, LWC renderers, Data Cloud grounding and Agent Ops governance, in production.