What can you build?
AutronAI separates data, AI capabilities, conversation, and deployment channels. Give each Agent only the data and tools its responsibility requires.
- Resources provide data, reusable instructions, and external tools.
- Agents perform a focused responsibility with selected resources.
- Chatbots combine Agents, Memory, and conversational configuration.
- Integrations deploy a Chatbot to Web, Messenger, Telegram, or Zalo.
- Analytics and Request History expose quality, latency, cost, and errors.
The model to remember
| Component | Responsibility | Attached to |
|---|---|---|
| Structured source / Documents | Provide data | Agent |
| Skill | Provide reusable instructions | Agent |
| MCP tool | Perform an external action | Tools Agent |
| Agent | Provide a focused AI capability | Chatbot |
| Memory | Provide durable context | Chatbot |
| Integration | Provide a communication channel | Chatbot |
Recommended setup order
- Choose one use case and its minimum data.
- Create or install the required Resources.
- Create an Agent and assign the correct Resources.
- Create a Chatbot and attach Agents and Memory in Chatbot Studio.
- Test expected, incomplete, and out-of-scope questions.
- Create an Integration only after the composition works.
- Monitor real requests and improve one layer at a time.
Security baseline
Plan a use case
- Describe the user, question, and desired outcome in one sentence.
- List required data and remove sensitive data that is not needed.
- Choose one primary Agent instead of composing many Agents immediately.
- Choose a test channel and an acceptance owner.
- Write at least five test questions before configuration.
First workspace checklist
- Entity names are distinct from production names.
- Resources contain only necessary data.
- The Agent has a clear responsibility and limits.
- The Chatbot has the intended Agent and Memory.
- The test Integration points to the correct Chatbot.
- Request History records successful and failed tests.