A ready-to-use chat interface for every agent you build.
Every agent in Bud AI Foundry comes with Agent Widget, an embeddable chat window. Set its design, behaviour and features, copy one snippet into your application, and your users can talk to the agent without your team building a chat UI.
From agent to embedded chat, in one walkthrough.
Preview the widget, configure its design, style, features and content, then copy the snippet into your application.
Skip building the chat UI. Embed the agent.
Agents deployed in a project could already be called through code snippets, but your team still had to design and build a chat interface on top of each one. Agent Widget ships that interface with the agent. If you have built an agent that automates HR operations, embed its widget in your HR portal, or wherever your employees already work.
Three steps from agent to embedded chat.
The widget editor shows the chat window on one side and its settings on the other, so every change is visible before your users see it.
-
01
Build the agent
Create and deploy the agent in a Bud AI Foundry project. Its widget comes with it.
-
02
Configure the widget
Set the launcher, chat panel and window size, then work through the Design, Style, Features and Content tabs while the live preview updates.
-
03
Copy the snippet
Copy the snippet code and add it to your application. Your users get a ready-to-use chat interface for the agent.
Fit the chat to your application and your users.
Six areas of settings decide how the widget looks, how conversations behave and what the assistant is allowed to do.
Layout and scenarios
Customise the launcher, the chat panel and the window size. Scenarios preview the chat across different conversation situations, so you see exactly what your users will see.
Design
Set the domains the widget may run on and the context it receives. Name the assistant, write the chat title and greeting, and add an avatar, your logo, colour themes and accents.
Conversation behaviour
From the Style tab, decide whether users can copy responses, give feedback, regenerate an answer or share a conversation.
How the agent shows its work
Choose how reasoning steps, tool calls, sources and plans are displayed, and set how suggestions and attachments appear.
Features
Configure human-in-the-loop and model selection, and turn attachments, tool usage, the audio player and generative UI on or off.
Content
Add suggested actions or queries tailored to the agent, plus a disclaimer or footer note.
You decide where it runs and what it can do.
The widget carries the agent into your application, so its limits are set in the same editor.
Allowed domains
In the Design tab, you set the domains where the widget is allowed to run, so it appears only in the applications you choose.
Capabilities, on or off
Human-in-the-loop settings, model selection, tool usage, attachments and generative UI are set per widget, to suit the use case.
Where Agent Widget fits.
Agent Widget is part of Bud AI Foundry, where the agents behind it are built, deployed and governed.
Product
Bud AI Foundry
Where the agent behind the widget lives
Inference, guardrails, observability, evaluations and cost control on one serving plane, with agents built and governed in the same place.
Explore Bud AI FoundryFeature
Skills
Give the agent reusable capabilities
Package instructions, workflows and capabilities once, have Bud SENTRY scan them, and attach them to the agents your widgets serve.
Explore SkillsPut an agent in front of your users.
Tell us where your users work and what the agent should do. We will show you the agent, its widget and the snippet that puts it in your application.