Agents Overview
What is an Agent?
An Agent in Vaillancore is your deployed AI assistant — trained with your system message, powered by your selected model, and optionally enhanced with Resources for deeper contextual understanding.
Agents are configurable via the Vaillancore Admin Dashboard and are the core interface between your end-users and the Vaillancore AI infrastructure.
Key Agent Concepts
Each Agent has:
| Field | Description |
|---|---|
| Name | A label you choose to identify the agent in your dashboard |
| Model | The AI model powering your agent (e.g., GPT-4.1 Mini) |
| System Message | An instruction you provide to guide your agent's tone, persona, or purpose |
| Resources | Optional knowledge bases (see Resources Overview) that enhance the agent's responses |
| Status | Active or disabled — determines if the agent is callable via your API |
Creating an Agent
From the Admin Dashboard, you can create a new Agent by:
- Navigating to the Agents tab
- Clicking "Create Agent"
- Entering a:
- Name
- System Message (see tips below)
- Model (if configurable)
- Attaching any Resources the agent should use
- Saving the agent
Once created, the agent is immediately usable via the Vaillancore /chat endpoint — just provide the correct agent_id.
The System Message
This is your agent's internal "character guide." It helps shape the behavior of responses — tone, style, and intent. It's not visible to end-users but heavily influences output.
Examples:
- "You are a professional support agent for a SaaS company. Always be friendly, concise, and helpful."
- "You are a technical assistant for our internal API documentation. Be thorough and give code examples when relevant."
Tip: Keep it short but specific. Test variations to dial in the tone.
Attaching Resources
Resources enhance your agent's performance by giving it access to domain-specific knowledge. When a user sends a message, the most relevant Embeddings (from attached Resources) are retrieved and injected into the prompt, giving the model more context to reason with.
To attach Resources:
- Open the Agent in the dashboard
- Scroll to the Attached Resources section
- Select one or more Resources to include
- Save changes
When to Use Multiple Agents
You can create as many agents as needed. Use multiple agents when:
- You want different tones or personas (e.g., "Friendly Bot" vs "Formal Bot")
- You want different knowledge scopes (e.g., Support Bot vs Product Demo Bot)
- You're tracking usage or metrics per deployment (e.g., separating bots for customers or teams)
Example Agents
| Name | Purpose | System Message (example) |
|---|---|---|
| OnboardingBot | Guides new hires through HR docs | "You're a friendly HR guide for new employees…" |
| SupportAI | Answers user product questions | "You are a concise, professional support agent…" |
| ComplianceHelper | Assists with regulatory questions | "Answer strictly from attached compliance policies…" |
Agent Status
You can toggle the status of any agent to active or inactive. Inactive agents cannot be queried through the API and will be excluded from usage-based billing.