Getting started with AgentHere
What AgentHere is, who it's for, and how to go from sign-up to your first agent reply in a couple of minutes.
AgentHere is a managed AI-agent platform. You assemble an agent from a template or from scratch, equip it with skills, and talk to it in a chat. Under the hood there's an OpenAI-compatible proxy, token metering, and card payments.
What's available right away
- Web chat on
agenthere.online— the primary way to work, no install required. - Desktop app for Windows, macOS, and Linux — for those who want an agent on the desktop.
- API and MCP server — to connect AgentHere to your own code, Cursor, Claude Desktop, or any OpenAI-compatible client.
Who it's for
The platform is built for people who want a task-tailored AI without fiddling with APIs:
- specialists and managers who want a ready assistant for their work;
- developers and IT pros — for code and routine;
- students and researchers;
- anyone who wants to try AI before signing up — anonymous chat is available on the home page.
Three steps to your first reply
1. Sign up
On the sign-up page enter an email and password, or sign in with Google. After you confirm your email, 1,000,000 tokens are credited to your account — a one-time free grant. See Billing & tokens for details.
2. Create an agent
Go to My agents and pick a template — a personal assistant, an analyst, an editor. A template already ships with a suitable system prompt and model settings; you can clone it straight into your account. How it works — in Create your first agent.
3. Start a conversation
Open the chat, choose an agent in the switcher at the top, and send a message. Replies stream in and the conversation is saved to history. More in Your first conversation.
What's next
- Learn how agents are built and what they consist of.
- Extend an agent with skills — ready-made Markdown playbooks.
- If you write code, start with the API reference.