Skip to main content
AgntLabs lets you build AI support agents that answer customer questions using your own knowledge base. Your agent learns from the files, websites, text and Q&A pairs you give it — then handles conversations on your website automatically. No coding required. A working agent takes a few minutes.

​
How it works

1

Create an agent

Give it a name, a role and a goal. It starts life as a draft — nothing is live until you publish.
2

Add knowledge

Upload files, paste text, add website URLs, or write Q&A pairs. Everything lands in one Knowledge base panel.
3

Publish

Publishing reads your sources, splits them up and embeds them. This is the step that turns a draft into a working agent.
4

Test, then deploy

Chat with your agent right beside its settings. When it answers well, copy one script tag onto your site.

​
What’s in the dashboard

Home

Conversations, messages, response time, leads and CSAT — plus how conversations ended.

Your Agents

Build and test an agent on one page: chat on the left, configuration on the right.

Conversations

Every conversation your agents have had, with transcripts, outcomes and captured leads.

Customize

How your agent sounds and looks — tone, greeting, colour, logo and position.

Analytics

Resolution rate, deflection, CSAT, and the questions your agent could not answer.

Actions

Let your agent raise tickets, book appointments, track orders and capture leads.

Channels

Connect your agent to Slack. WhatsApp is coming.

Deploy

One script tag. Guides for WordPress, Shopify, Wix, Squarespace and React.

​
Who it’s for

Businesses that want to automate support without dropping the quality. If your team answers the same questions every week — shipping, returns, pricing, how something works — your agent can take those.

​
What your agent will not do

It answers from your knowledge base and nothing else. Ask it something outside your material and it says so rather than inventing an answer. Those moments are recorded, so you can see exactly which questions your agent could not handle and fix them.

​
Next

Follow the Quickstart to get an agent live.