Quickstart: your first agent in ten minutes

This guide takes you from a new account to an agent that answers from your own document, first in the Playground and then from a terminal through the API.

Before you start

• An API key from OpenAI, Anthropic, Google Gemini or Groq. Agents always run on your own provider key.
• A PDF the agent should answer from, such as a returns policy or a product guide (up to 50 MB).
• A terminal with curl for the last step.

Steps

1. 1

   Create your account

   Go to app.sentientone.ai and sign up with email and password, Google or Microsoft. You start with 200 free credits.

2. 2

   Add your LLM key

   Agents run on your own model provider key; there's no shared one, so nothing answers until you add it. Create a key with OpenAI, Anthropic, Google Gemini or Groq, then in the sidebar open Configure → LLM keys → Add key. It's encrypted with AES-256-GCM and only ever shown masked. See Add your LLM key.

3. 3

   Create an agent

   Open Agents and choose New agent. Pick a template under Start from or start from Custom, give the agent a name, choose a provider and model, pick the LLM key you just saved (or paste one) and write the system prompt. Select Create agent.

4. 4

   Add knowledge

   On the agent's Knowledge tab, open Documents and upload your PDF. Each page costs 5 credits, charged once. When the status shows Ready, the agent searches it on every question. There is no switch to turn knowledge on; it is used as soon as the agent has indexed content.

5. 5

   Test it in the Playground

   Open Playground, pick the agent and ask a question your document answers. Each reply shows its latency, tokens, sources and any tool calls. Each reply costs 1 credit.

6. 6

   Copy your API key and the agent id

   Open API keys in the sidebar. Copy your secret key (it starts with sk-so-) and the agent's id. Keep the secret key on your server: never put it in a web page or a mobile app.

7. 7

   Call the API

   Send a message to the streaming endpoint. -N turns off curl's buffering so you see the events as they arrive.

   export SENTIENTONE_API_KEY="sk-so-..."export AGENT_ID="your-agent-uuid"curl -N https://api.sentientone.ai/v1/chat/stream \  -H "Content-Type: application/json" \  -H "X-Api-Key: $SENTIENTONE_API_KEY" \  -H "X-Agent-Id: $AGENT_ID" \  -d '{"message": "Can I return an item I have worn once?"}'

8. 8

   Read the events

   The answer arrives as data: lines: meta first (with the conversation_id and trace_id), then sources, then delta pieces of text, and done with token usage. Send the conversation_id back with the next message to continue the conversation. See Streaming events.

Where to go next

• Put the agent on your website with the widget.
• Give it your systems as tools with MCP.
• Automate a process around it with a workflow.
• Handle errors properly: Errors lists every code.

If something goes wrong