Skip to main content
SecureChat Agent Configuration

​
Creating a SecureChat Agent

SecureChat is not a standalone product, but a type of intelligent agent that can be created and configured within Prisme.ai, using either AI Store or AI Knowledge, depending on the user’s permissions and organizational context. It offers a secure, AI-powered conversational experience that can be public to the instance or scoped to a specific entity, based on access rules and SSI (Security & Information Systems) configurations.

​
Creation Options

You can create a SecureChat agent using one of the following methods:
Users with sufficient permissions can create SecureChat agents manually through the platform UI:
  • Navigate to AI Store (for general-purpose assistants)
  • Or go to AI Knowledge > Agents (for knowledge-centric assistants)
  • Set visibility (public, group-limited on Users tab, etc.)
  • Define agent prompt, tools, documents, and runtime settings
Use automation to define and deploy agents across environments:Example workflow:
slug: create-agent-securechat
name: Create agent SecureChat
do:
  - set:
      name: agents
      value:
        - name: SecureChat
          category: Productivity
          public: true
          description: Your AI-powered professional assistant for daily work tasks.
          ai:
            max_tokens: 1500
            prompt: |
              You are a virtual assistant designed to support employees in their daily professional tasks...
  - repeat:
      on: '{{agents}}'
      do:
        - Knowledge Client.Projects - Update or Create project:
            data: '{{item}}'
This ensures traceability and multi-environment management.

​
Agent Capabilities

A SecureChat agent typically supports:
  • Summarization of documents and meeting notes
  • Comparison between files or texts
  • Rephrasing and rewriting content
  • Drafting emails, notes, or reports
  • Web search (if enabled) with source citations
  • Secure document handling with contextual processing
Each response type is designed with formatting standards and structured output (e.g., tables, bullets, or paragraph summaries).

​
Configuration Options

  • Set public: true to make the agent visible to all users.
  • Or public: false and then assign access by group, entity, or workspace role using an automation on AI Builder
  • Connect to multiple LLMs (OpenAI, Mistral, Claude, Ollama…)
  • Define a prompt depending on use case to define tone, role, behavior

​
Optional Tools & Webhooks

Agents can include tools and hooks to enhance capabilities:

Tools Integration

Add tools to your SecureChat agent:
  • Web browsing
  • Code Interpreter
  • Image Generation
  • Custom tools

Webhooks

Use webhooks for:
  • Pre-processing messages before LLM call
  • Validating user input or controlling access
  • Scanning and classifying uploaded documents
  • …

​
Monitoring Agent Activity

You can monitor the SecureChat agent’s activity through:
  • The Activity tab inside the agent’s workspace (AI Knowledge)
  • The AI Builder interface, if the agent was deployed via automation
  • Optionally, activity logs can be streamed to ElasticSearch/OpenSearch
Tracked events include:
  • Agent usage
  • Chat events
  • Feedback collection
  • RAG/document interaction logs

​
Next Steps

AI Store

Learn how to manage and deploy agents in the Store

AI Knowledge

Organize collections and document-based agents

Monitoring & Logging

Track usage and agent activity in real-time

Create an agent

Learn more about agent creation