TackUI

Open Source · AG-UI Protocol

Your own AI agents — built, installed, or connected

Build an agent without code, install a ready-made one with a click, or connect your own backend — all in one ChatGPT-style chat that's actually yours.

Get started — free Self-host on GitHub

Elastic License 2.0 · No credit card required

TackUI — Unified agent frontend

Research Agent

Analyze the Q3 market trends for SaaS tools

I'll research Q3 SaaS market trends. Let me gather data from multiple sources.

🔍web_search✓ done

"Q3 2024 SaaS market trends"

Based on the results, the SaaS market in Q3 showed strong growth in AI-native tools

Message Research Agent…

Three ways to get an agent

Build one, install one, or connect one — they all end up in the same place: your agents, in one chat.

Build it

Open the builder, pick a model, add the tools and connectors it needs. No code, no backend — your agent shows up in the sidebar.

GPT-4o

web_searchrun_code+ Add tool

Connect an MCP server

Build your first agent →

Install it

Ready-made agents from the catalog — add one to your workspace in a single click. No endpoints, no config.

Release Notes Writer

Merged PRs → clean changelogs

Add

Meeting Notes Agent

Transcripts → action items

Add

Browse the catalog →

Connect it

Already have an agent backend? Point any AG-UI-compatible endpoint at the registry — it gets the full frontend for free.

{

"name": "Research Agent",

"kind": "agui",

"endpoint": "https://…/agent"

}

AG-UI quickstart →

Agent builder

An agent of your own, in five minutes

Name it, pick a model, choose the tools it can use, wire up an MCP connector or two. No code, no backend — the builder turns your description into a working agent that appears in your sidebar.

Start building

TackUI — Agent builder

Name

Release Notes Writer

Model

GPT-4o· balanced

Tools

get_pull_requestlist_commitsrun_code+ Add tool

Connectors

GitHub MCPconnected

+ Connect an MCP server

Create agent

Agent catalog

Or skip the setup entirely

Ready-made agents from the catalog — add one to your workspace with a single click. No endpoints, no config. Install, open a chat, go.

Browse the catalog

TackUI — Agent catalog

Search agents…

Release Notes Writer

Merged PRs → clean changelogs

Add

Meeting Notes Agent

Transcripts → action items

Add

Inbox Triage

Sorts, labels, drafts replies

Add

Competitive Scan

Tracks rivals, weekly digest

Add

Whatever their origin — one chat

Built, installed, or connected — every agent lands in the same sidebar and the same chat.

Every agent, one chat

Built, installed, or connected — every agent lives in the same sidebar. Switch agents per conversation; the interface never changes.

Research AgentCode AssistantData Analyst

Human-in-the-loop approvals

Agents can pause and ask for your sign-off before doing anything sensitive — you approve right inside the chat.

See the work as it happens

Watch agents call tools live — structured, readable renders instead of raw JSON.

web_search✓ completed

"Q3 2024 SaaS market trends analysis"

Conversations that persist

Every chat is a saved thread per agent. Close the tab, come back tomorrow — the history is waiting.

For developers

Bring your own backend

TackUI speaks the AG-UI protocol natively — built on CopilotKit, the 1st-party AG-UI client. Any backend that speaks AG-UI gets the full frontend: streaming, interrupts, tool renders, persisted threads.

Any framework — LangGraph, Agno, or your own server
Everything standard — streaming, interrupts, and tool renders come from the protocol, not custom code
Self-host it — Elastic License 2.0, unlimited everything, your infrastructure

Self-host on GitHub →Read the AG-UI docs →

{

"name": "Research Agent",

"kind": "agui",

"endpoint": "https://…/agent"

}

Built on open standards

AG-UI ProtocolCopilotKitBetter AuthNext.js 16

Simple pricing

Free to start. Upgrade when you need more agents or team seats.

See plans

Start building with TackUI today

Build your first agent in minutes — no code required.

Get started — free