Tisane

Tisane is a powerful and versatile Natural Language Processing (NLP) solution, often described as a 'Swiss Army knife' for user-generated content. Developed by Tisane Labs, it provides a robust API for automatic content moderation, text analysis, and data extraction. The platform is engineered to understand the nuances of online communication, with a special focus on what it calls “dark NLP”—language that is full of errors, slang, obfuscation, and adversarial text. Tisane goes beyond simple keyword filtering to offer deep contextual understanding, making it an essential tool for maintaining safe and clean online environments.

The core philosophy of Tisane is to combine automation with human ingenuity, providing tools that are versatile, neutral, and forward-thinking. It is designed to be highly flexible, offering SaaS, on-premise, and embedded deployment options to fit various infrastructure and privacy requirements. This makes it suitable for a wide range of applications, from large-scale social platforms to specialized forensic software.

How to use Tisane

Using Tisane is straightforward for developers. The primary method of interaction is through its REST API. Here's a typical workflow:

1. Sign Up: Choose a subscription plan on the Tisane website, starting with the free 'Community' tier to get started.
2. Get API Key: After signing up, you will receive an API key to authenticate your requests.
3. Integrate the API: Send text content to the Tisane API endpoint via a POST request. The content can be a simple string or a more complex object. You can specify the language or let Tisane's language identification feature handle it automatically.
4. Process the Response: The API returns a detailed JSON object containing the analysis. This includes detected issues like abuse or problematic content, with explanations, severity levels, and specific offsets in the text. The response also includes extracted entities, topic models, and other linguistic data.
5. Deploy Your Way: For different needs, you can connect to the public SaaS cloud, deploy Tisane on your own servers (On-Prem), or embed it directly into your desktop or server application for offline processing.

Core Features of Tisane

• Advanced Content Moderation: Detects a wide range of problematic content, including cyberbullying, hate speech, personal attacks, sexual advances, adult topics, criminal activity, and profanity.
• Explainability: Unlike many black-box AI models, Tisane provides clear explanations for why a piece of content was flagged, enhancing transparency and aiding human moderators.
• Algospeak & Obfuscation Detection: Uniquely capable of understanding and flagging 'algospeak' (e.g., 'unalive' instead of 'kill') and other adversarial text manipulations designed to bypass traditional filters.
• Entity Extraction & Topic Modeling: Identifies and extracts named entities (people, places, organizations) and performs granular topic modeling to understand the subject matter of the text.
• Built-in Multilingual Support: Natively supports over 35 languages with built-in automatic translation, allowing for consistent moderation across a global user base.
• Flexible Deployment: Available as a cloud-based SaaS, a self-hosted on-premise solution for maximum privacy and control, and an embeddable library for native applications.

Use Cases for Tisane

Tisane's versatility makes it suitable for various industries:

• Online Communities & Social Media: Protect users from harassment, hate speech, and toxic behavior to foster a safe and engaging community.
• Gaming: Safeguard gaming communities by moderating in-game chat, player usernames, and forum discussions in real-time. It includes a special mode for game chat processing.
• E-commerce & Marketplaces: Prevent review bombing, filter out illicit merchandise listings, and stop covert promotions to maintain the integrity of the platform.
• Law Enforcement & National Security: Embedded in major investigation and forensic suites, Tisane helps agencies analyze digital evidence and communications, even on legacy hardware without Java dependencies.
• Generative AI Moderation: Acts as a trust and safety layer for applications using large language models (LLMs) like ChatGPT or Bard, ensuring the generated content is appropriate and safe.

Advantages of Tisane

Tisane stands out from other moderation tools due to several key advantages:

• Deep Linguistic Analysis: Its focus on "dark NLP" means it excels at understanding messy, real-world user content that other systems might miss.
• High Performance & Lightweight: The solution is built with native libraries for Linux, Windows, and more, ensuring it runs efficiently even on legacy hardware without requiring heavy dependencies like Java.
• Neutrality by Design: The tool's purpose is to detect and classify content based on defined rules and patterns, not to make subjective judgments, which reduces bias.
• Comprehensive Solution: It's more than just a profanity filter; it's an all-in-one text analysis suite that provides moderation, entity extraction, and topic modeling in a single API call.

Pricing and Plans

Tisane offers a freemium pricing model with several tiers to accommodate different scales of operation:

• Community Plan: $0/month. Includes 50,000 requests per month with a rate limit of 10 hits per minute. Ideal for small projects and testing.
• Small Plan: $99/month. Includes 100,000 requests per month with a rate limit of 120 hits per minute and limited technical support.
• Medium Plan: $159/month. Includes 400,000 requests per month, a 120 hits per minute rate limit, full technical support, and CORS support.
• Large Plan: $899/month. Includes 2,500,000 requests per month, a 2000 hits per minute rate limit, integration assistance, and limited customization.
• On-Premise/Enterprise: Custom pricing is available for users with high volume needs, strict privacy requirements, or the need to run Tisane in an air-gapped environment.