Imperative · Generative UI

AG-UI Control Demo

AI agents should not stop at paragraphs. AG-UI (Agent UI) lets them render buttons, forms, and data surfaces in real time — you already see Layer 1 on every virgent.ai page via our site assistant.

This sandbox shows Layer 3: full canvas control for enterprise workflows. Same philosophy, bigger surface area.

Pair with: A2UI Protocol Demo →

GenUI & AG-UI services →

The problem AG-UI solves: one rigid UI cannot fit every person. A first-time visitor, a power user, and someone with low vision all need a different layout. Here the agent reshapes the same app to match each user’s intent and ability — no redesign, no redeploy, no extra “modes” to maintain.

Northwind — live product UI

AG-UI Runtime

Ask the agent to help

These are real user struggles. The agent flexes the same app to fit each one — no redesign, no redeploy.

Tap a request below, or type one — watch the live app on the right reshape itself.

Northwind

SearchAlertsCartSettingsHelpSign up

Summer sale

Save 30% on annual plans this week only

Welcome back to Northwind

Track orders, manage your team, and explore new features across your workspace.

Recent orders

Tap to view details

Saved items

Tap to view details

Team activity

Tap to view details

Usage report

Tap to view details

Newsletter

Get weekly product tips in your inbox.

Recommended

Connect your CRM
Invite a teammate
Set up alerts

Business use case

What this demonstrates: AG-UI in two places: our site chat generates navigation buttons and in-chat forms from every conversation (Layer 1–2). This demo shows Layer 3 — agents that spawn full forms, tables, and charts on a live canvas when the task demands it.

How it relates to your business: Stop losing visitors between answer and action. Route them with generated buttons. Capture them with inline forms. For complex products, let the agent compose the right surface per user — without a redesign sprint for every persona.

Pain it solves: Chat-only AI creates a wall of text and a navigation hunt. Static UIs create ten versions nobody maintains. Neither scales when every user needs a different next step.

Who it is for: Product, UX, and growth leaders who want agents that guide, capture, and adapt — not chatbots that dump paragraphs and hope.

Real-world examples:

A low-vision user says “I can’t find the login button” — the agent enlarges sign-in, dims the clutter, and can scale up all text. Real-time accessibility, no separate mode to maintain.
A data team says “add a revenue KPI” then “create a low-inventory alert” — the agent composes new dashboard widgets and automations on the fly, no ticket or deploy.
A founder says “rebrand to teal and round the corners” — the agent edits design tokens and the entire product restyles live from one source of truth.
Ops manager says “create a pipeline table” then “add Maria to the table” — the agent edits the live grid, no developer deploy.
Site visitor asks about RAG — assistant answers and renders “Explore RAG Services”, “Try Live Demos”, and “Book a Call” buttons (live on virgent.ai today).

Generative UI, safely deployed

We wire agents to your design system with controls, guardrails, and human-in-the-loop.

Explore AG‑UI options

Where AG-UI shows up at Virgent

From the marketing site to client deliverables

In production today

• Site assistant: dynamic action buttons on every response
• In-chat capture: booking and deck request flows
• Thinking state: suggestion chips while the model reasons
• RAG-backed routing: links matched to services and case studies

What we build for clients

• Adaptive dashboards: widgets appear by role and task
• Dynamic forms: fields add/remove based on answers
• Agent-driven ops UIs: tables, charts, approvals inline
• Design-system safe: guardrails + human-in-the-loop