Ejento AI
Guides
QuickstartRecipesREST APIsRelease NotesFAQs
Guides
QuickstartRecipesREST APIsRelease NotesFAQs
Ejento AI
  1. Guides
  • Basic Operations
    • Features
      • Organization → Projects → Assistants → Teams Hierarchy
    • Guides
      • Login/Signup
  • Assistants
    • Overview
    • Features
      • Assistant Access Control
      • Assistant Evaluation
      • Evaluation Metrics
      • URL-based Chat Thread Creation and Prepopulation
      • Reasoning Patterns
      • Staging & Publishing
    • Guides
      • Add Assistant
      • Evaluate Assistant
      • Edit Assistant
      • Assistant Edit Access
      • Embed Assistant
      • Delete Assistant
      • Add Favourite Assistants
      • View Assistant Id
      • Voice Calling with Assistants
  • Corpus
    • Overview
    • Features
      • Corpus Permissions
      • PII Redaction
      • ETag Setup for Corpus Incremental Refresh
    • Guides
      • Assistant Corpus Setup
      • Assistant Corpus Settings
      • Corpus Access Control
      • Corpus Connections
      • View Corpus Id
      • View Document Id
      • Tagging
        • Corpus tagging
        • Document tagging
  • Teams
    • Overview
    • Guides
      • Add a Team
      • Edit a Team
      • Delete a Team
  • Projects
    • Overview
    • Guides
      • Add a Project
      • Edit a Project
      • Managing Assistants in a Project
      • Delete a Project
      • View Project Id
  • User Settings
    • Overview
    • Features
      • Ejento AI User Access Levels
    • Guides
      • Add new user
      • Invite Users via Link or Email
      • View my User Id
  • API Keys
    • Overview
    • Guides
      • How to generate API Key and Auth Token
  • Workflows
    • Overview
    • Features
      • Workflow Access Control
    • Guides
      • Add Workflow
      • Workflow Chat
      • Workflow Edit Access
  • Tools
    • Overview
    • Guides
      • Tools Overview
      • Create External Tool
      • Connect Tool to Assistant
  • Analytics
    • Overview
    • Guides
      • Analyzing Data in the Analytics Dashboard
  • Chatlogs
    • Overview
    • Guides
      • Managing Chatlogs
      • View Chatlog & Chat thread Id
  • Integrations
    • Overview
    • Guides
      • Email Indexing
      • Microsoft Teams
      • Sharepoint Indexing
      • Google Drive Connector
      • MS Teams Integration Setup
      • Creating a Connection in Credential Manager
      • Slack App
      • Discord Bot
  • Ejento AI Shield
    • Overview
    • Features
      • Understanding Guardrails
    • Guides
      • How to enable Guardrails
  • Assistant Security
    • Overview
    • Features
      • Assistant Red Teaming
    • Guides
      • Red Team an Assistant
  • Permission Sets
    • Overview
    • Guides
      • Add Permission Set
  • Usage
    • Overview
    • Guides
      • Set Spend Limits
  1. Guides

Ejento AI Shield

Ejento AI Shield is a comprehensive safety and governance layer designed to protect your AI assistants from misuse, hallucinations, data leakage, and policy violations. It provides configurable guardrails that ensure AI behavior remains secure, ethical, accurate, and aligned with organizational boundaries. You turn guardrails on per assistant, so each assistant gets only the protection it needs. Explore the features and guides below.

What you'll learn#

What each guardrail does: the five guardrails, and the kinds of risk each one protects against.
How to turn guardrails on: switch them on in the assistant's Manage panel, under Behavior > Guardrails.
How to set up Topical Control: write guidelines that define what the assistant may and may not discuss.
How to set up PII Redaction: choose which kinds of personal information to hide, and which words to never redact.
How to choose the right mix: which guardrails fit a public chatbot, an enterprise assistant, or a data-sensitive workflow.

How guardrails work#

Ejento AI has five guardrails. Three work as soon as you turn them on. Two need a short setup.
GuardrailWhat it doesSettings
Topical ControlControls what topics the assistant can discuss, based on guidelines you write.Your guidelines, in Edit guidelines
Ethical ModerationKeeps responses ethically sound, for example by refusing harmful or illegal requests.None, on or off
Jailbreak DetectionStops attempts to trick the assistant into ignoring its rules.None, on or off
HallucinationReduces made-up information in responses.None, on or off
PII RedactionHides personal information, such as names, phone numbers, and email addresses.Categories to redact and Never redact, in View settings
Guardrails apply after you deploy
You change guardrails in the assistant's Staging environment. People only get the protection after you click Deploy at the bottom of the Guardrails panel.

Get started#

1.
Pick the guardrails you need for your use case. See Understanding Guardrails.
2.
Open your assistant and switch to Staging in the Manage panel. See How to Enable Guardrails.
3.
Turn on each guardrail under Behavior > Guardrails. See How to Enable Guardrails.
4.
Set up Topical Control and PII Redaction, if you turned them on. See How to Enable Guardrails.
5.
Deploy the change so the guardrails protect the live assistant. See How to Enable Guardrails.

Features#

Understanding Guardrails
Explore the different types of guardrails in Ejento AI, including Ethical Moderation, Jailbreak Detection, Topic Control, Hallucination Detection, and Sensitive Data Masking.

Guides#

How to Enable Guardrails
Step-by-step instructions to configure and activate AI Shield guardrails for your assistants, ensuring safe and controlled AI interactions.

Where to go next#

Assistant Security: red team an assistant to test how well your guardrails hold up.
Assistants: create, set up, share, and deploy AI assistants.
Tools: choose which tools each assistant can use.
Previous
Discord Bot
Next
Overview