Custom AI agents,
trained on your content.
Import your documents and websites, get grounded answers with citations, and embed the agent anywhere — on your site, in your product, or through the API.
No credit card required · Free forever plan · 2-minute setup
Trusted by 12,400+ builders
Employees receive 20 days of paid time off (PTO) per year.
[1] PTO Policy12,400+
Agents built
98.2%
Answer grounded in sources
1.2s
Average time to index
24ms
Retrieval latency
Powering agents at teams like
Everything you need to ship an agent
From first document to production widget — the whole pipeline is built in.
Train on anything
Upload PDFs, DOCX, Markdown, or CSV. Paste text. Point at a URL or crawl a whole sitemap. Everything gets indexed into your agent's private knowledge base.
Grounded, cited answers
Every reply is retrieved from your own content and cites the exact source it came from. No invented facts, no confident guessing.
Embed in one tag
Drop a single script snippet on your site and get a fully branded chat bubble backed by your agent. It works on any stack.
An API that just works
Scoped API keys, a clean REST interface, and streaming responses. Connect your agent to any product, bot, or internal tool.
Fast by default
Chunking, indexing and retrieval are tuned for production traffic, so your agent answers in the time a customer expects.
Know what it's doing
Usage analytics, message volume, popular questions, and per-project health — so you can see exactly how your agent performs.
Three steps to a working agent
No prompt engineering, no vector database, no infrastructure.
Create your agent
Name it, describe what it should know, and set its personality. Takes about thirty seconds.
Bring your content
Upload documents, paste knowledge, or connect a site. We index and chunk everything automatically.
Chat, embed, ship
Test it in the builder, embed the widget on your site, or call it from your own code.
Your agent, in your code
Create a key, call the endpoint, get a grounded answer. That's the whole integration.
- Scoped, revocable API keys
- Streaming and non-streaming responses
- Citations included in every payload
- Per-project isolation by default
curl -X POST https://app.agentsgpt.ai/api/v1/projects/{id}/chat
-H "Authorization: Bearer $AGPT_KEY"
-H "Content-Type: application/json"
-d '{"message": "What's our refund policy?"}// response
"answer": "All standard plans include a 30-day..."
"citations": [{ "documentName": "Refund Policy", "index": 1 }]
Teams ship faster with AgentoGPT
“We replaced a 40-page help centre with a support agent that cites its sources. New teammates get answers in seconds now.”
Priya Raman
Head of Support, Northwind
“The API is the part that sold us. Dropped it into our product in an afternoon, and the grounding is genuinely better than the big providers.”
Marcus Feld
CTO, Lumen Labs
“Our legal team approves every answer because it shows its sources. That was the blocker for every other tool we tried.”
Sofia Almeida
Head of Ops, Kestrel
Simple plans that scale with you
Start free. Upgrade when your agent earns its keep.
Free
Explore what AgentoGPT can do.
- 1 project
- 5 sources per project
- 100 messages / month
- Chat with citations
- Community support
Starter
For solo builders shipping a first agent.
$360 Billed $180 yearly
- 10 projects
- 50 sources per project
- 5,000 messages / month
- Embeddable chat widget
- REST API access
- 5 API keys
Pro
For teams running agents in production.
$1512 Billed $756 yearly
- 50 projects
- 500 sources per project
- 50,000 messages / month
- Custom widget branding
- Full REST API + API keys
- 15 team members
- Priority support
Enterprise
For teams with scale and compliance needs.
$5736 Billed $2868 yearly
- Unlimited projects & sources
- Unlimited messages
- Unlimited team members
- SSO / SAML
- Dedicated support & SLA
- On-prem deployment options
Frequently asked questions
How does the agent avoid making things up?
Every question is answered from chunks retrieved out of your own knowledge base. If the answer isn't there, the agent says so instead of inventing one — and always shows which source each claim came from.
What file types can I upload?
PDF, DOCX, TXT, Markdown, HTML, CSV, JSON and XML. You can also paste raw text, point at any public URL, or hand us a sitemap.xml and we'll crawl every page it lists.
Can I use this on my own website?
Yes. Add one script tag and a chat bubble appears, styled with your project's brand colour. It works with any CMS or framework — there's no build step.
Is there an API?
Every plan includes REST access. Create a scoped API key from your profile and call your agent for chat, projects, and documents from any application.
How is my data handled?
Your content belongs to you. It's stored per-project, never used to train shared models, and is only ever retrieved for the person or app that asked the question.
What happens when I hit my message limit?
Your agent keeps working — you just won't be able to start new conversations until the period resets, or until you move to a plan with more room. Nothing is ever deleted.
Ready to build your first agent?
Free forever plan. No credit card. Your first agent can be answering questions in two minutes.