Version: 2.0

Create an Agent

This section provides a step-by-step procedure for creating a custom AI agent in the Vectara platform. Agents are configurable AI entities that can leverage models, instructions, and tools to handle queries intelligently. You can create agents with the Vectara Console or programmatically through the API. The Console is ideal for quick setups, while the API suits automation or advanced integrations.

Prerequisites

Access to a Vectara account with appropriate permissions.
For RAG tool usage: At least one populated corpus (uploaded documents or data sources).
Familiarity with JSON for overrides and parameters.
Supported models ("gpt-4o", "claude-sonnet-4").

The agent creation process follows a linear flow in the UI:

Access the Agents page from the sidebar.
Click Add Agent.
General: Define initial information about the agent.
- Enter a Name like My Custom Agent or Chatbot.
- Add a Description like "Handles queries on internal data".
Model: Choose the LLM that you want to associate with this agent.
- Select a Model name such as gpt-4o.
- Configure Parameters in JSON for response control. For example:
  { "temperature": 0.5, "max_completion_tokens": 1024 }
Instructions: Set the behavior for the agent
- Add at least one Instruction like “Respond concisely and use tools when relevant. Prioritize RAG for internal knowledge”
- Optionally add more instructions from the library.
Tools: Enable capabilities for the agent.
- Add tools like Corpus Search tol_corpus_search and Web Search tol_web_search
- For each tool, set Input overrides in JSON.
  { "search": { "corpora": [ { "corpus_key": "my-corpus" } ] } }
Advanced: Fine-tune the agent with custom metadata.
- Add Metadata in JSON.
  { "team": "support" } or { "team": "engineering" }
- Click Create Agent to finish.