Build your first AI support agent
Triggerflow team · · 2 min read
In this guide you build a support agent that answers questions in a chat, remembers the conversation, and can look things up through an API.
1. Create the workflow
From the Overview, choose Create workflow and give it a name, for example Support agent.
2. Start with a Chat Trigger
Add a Chat Trigger. It starts the workflow every time someone sends a message, and passes the message on as chatInput.
3. Add the AI Agent
Connect an AI Agent after the trigger. In its options, describe the job in the system message, for example:
You are the support assistant for our shop. Answer briefly and politely. If you don't know, say so and offer to pass the question to a person.
4. Give it a model, memory and a tool
The agent has three connection points underneath:
- Chat Model: add an OpenAI Chat Model or Anthropic Chat Model and choose your credential. You can also use OpenRouter, Mistral, DeepSeek or a local Ollama model.
- Memory: add Simple Memory so the agent remembers earlier messages in the same chat.
- Tool: add an HTTP Request Tool pointing at your order or product API, and describe in plain words what it does. The agent calls it when a question needs it.
5. Test it in the editor
Select Open chat next to Execute workflow and send a message. Each answer is a run you can inspect: the prompt, which tools were called and what they returned.
6. Put it on your website
Open the Chat Trigger and turn on Make chat publicly available. Set a title and a welcome message, then activate the workflow. Triggerflow shows the link to your hosted chat page and the script to embed the chat widget on your site.