Guardrails

Visit Tool

Guardrails is an open-source toolkit for adding programmable guardrails to LLM-based conversational systems. It helps developers build trustworthy, safe, and secure AI applications by controlling LLM output.

Claim this tool

No Views Yet

At a glance

Pricing

Open Source

Free tier

Yes

API

Yes

Skill level

Technical

About

What is Guardrails?

NeMo Guardrails is an open-source toolkit designed to easily add programmable guardrails to LLM-based conversational applications. These guardrails control the output of large language models, allowing developers to define specific behaviors such as preventing discussions on unwanted topics, enforcing predefined dialog paths, or extracting structured data. The toolkit supports various LLMs like OpenAI GPT-3.5, GPT-4, LLaMa-2, and Falcon. It offers five main types of guardrails: input, dialog, retrieval, execution, and output, which can reject or alter data at different stages. NeMo Guardrails also provides mechanisms to protect against LLM vulnerabilities like jailbreaks and prompt injections, making it ideal for building trustworthy and secure AI systems.

Best used for

Ideal for developers who need to ensure LLM-based applications are safe, trustworthy, and adhere to specific conversational policies. Especially valuable for enforcing fact-checking in RAG systems, keeping domain-specific assistants on topic, and protecting against LLM vulnerabilities like jailbreaks and prompt injections.

Common actions

control LLM behavior

secure AI applications

define conversational flows

integrate guardrails

deepfakeopen-sourceautomated workflowcollaborationlow-code/no-codeworkflows"AI Agents"face swappinggithub copilot

Capabilities

Key features

Programmable guardrails
LLM vulnerability protection
Input/dialog/output rails
Colang modeling language
LangChain integration
Guardrails server

Target Audience

developerdata scientistproduct manager

Integrations

langchain

Pricing & Plans

Open Source

Free

FAQs

What types of guardrails can be implemented with NeMo Guardrails?

NeMo Guardrails supports five main types: input rails (for user input), dialog rails (for prompting LLMs), retrieval rails (for RAG chunks), execution rails (for custom actions), and output rails (for LLM-generated output). These can reject or alter data at various stages.

What is Colang and how is it used in NeMo Guardrails?

Colang is a modeling language specifically created for designing flexible, yet controllable, dialogue flows within NeMo Guardrails. It has a Python-like syntax and is used to define various types of guardrails, including conversational paths and moderation rules.

Can NeMo Guardrails be used with different Large Language Models?

Yes, NeMo Guardrails is designed to be compatible with multiple LLMs. It supports popular models such as OpenAI's GPT-3.5 and GPT-4, as well as open-source models like LLaMa-2, Falcon, Vicuna, and Mosaic, offering flexibility for developers.

Trending

Subcategories trending in AI Agents & Automation

AI Frameworks & Infra General-Purpose Agents Workflow Agents Personal Assistants RAG & Document AI Voice Agents

Trending

Explore

Browse AI tools by category

Content & Design Productivity & Business Coding & Development AI Agents & Automation Research & Education Wellness & Lifestyle Career Development Marketing & Growth Data & Analytics Customer Support & CX Finance E-commerce