Zentropi

What Zentropi does

Zentropi is a private AI infrastructure company that builds custom content labeling systems for text and images. The company powers its platform with CoPE-B (Contextual Policy Engine), a proprietary 4B active-parameter small language model purpose-built for content classification and released as an open-weight model on HuggingFace. CoPE-B claims frontier-model accuracy at less than 1% the size of standard LLMs and uses a training methodology designed to follow user-specified policies rather than the model's own priors, enabling dynamic policy changes without retraining.

Zentropi's product surface consists of a web-based no-code labeler builder, a Build API for creating and benchmarking labelers, a Run API for production classification, and an AI coding-agent skill (npm-installable) that integrates with Claude Code, Codex, Hermes, and OpenClaw so users can build and deploy labelers via natural language. The company supports both cloud API access and free self-hosting of the CoPE-B model for infrastructure sovereignty. Use cases marketed include feed ranking, content moderation, community management, AI guardrails, parental controls, and inventory management.

Zentropi monetizes through a freemium model: free basic labeler creation and deployment on the web, with paid subscription tiers for advanced capabilities and usage-based pricing for production API calls. The company is structured as a public benefit corporation, has 1-10 employees, and serves customers including Character.AI, Runway, Stanford, Koko, and Cantina across AI platforms, academic research, and community moderation use cases. The company was co-founded by Samidh Chakrabarti.

Name

Zentropi

Legal name

Zentropi

Website

https://zentropi.ai

Company type

Private

Operating status

Operating

Headcount range

1–10 employees

Short description

Zentropi is a private AI company that provides a platform for building and deploying custom content labelers (AI guardrails) for text and image classification, powered by its proprietary 4B-parameter open-weight model CoPE-B, sold via freemium, subscription, and usage-based API pricing.

Ownership category

akta.pro rank

Product category

AI Content Classification & Moderation Software

NAICS

Software Publishers (5132)

SIC

Services-Prepackaged Software (7372)

akta.pro primary industry

Model Development & Training Platforms (AutoML, Notebooks, Feature Stores) (HDAEANAB)

akta.pro secondary industries

Foundation Model Developers (LLM/Multimodal Model Labs) (HDAAACAA), AI Application Enablement Platforms (Copilot/Agent Frameworks, SDKs) (HDAEANAJ)

Keywords

• Content moderation software
• AI content classification
• Custom AI guardrails
• Content labeling platform
• Policy enforcement APIs

Headquarters

HQ city

Palo Alto

HQ country

United States

HQ region

North America

Markets served

GTM type

B2B

Offering type

Software

Cost components

Technology or R&D, Personnel, Infrastructure, Operations, Marketing or Sales

Revenue model

1. API Access (Cloud): Cloud-based API access for building, testing, and running labelers with subscription tiers for advanced capabilities
2. Usage-based API Calls: Pay-per-use model for API calls to Run API for classifying text or images
3. Freemium Tier: Free access to basic features with 'No subscription required' messaging; advanced capabilities require paid subscription

Pricing tiers

Go-to-market motion3 records

Distribution channels4 records

Marketing channels4 records

Core offering

Zentropi provides a software platform for instantly building, testing, and deploying custom content labelers (AI guardrails) that classify text and images against user-specified rules. The platform is accessible through a no-code web interface, Build and Run REST APIs, and an AI coding agent skill, and is powered by its proprietary CoPE-B 4B-parameter small language model that is available open-weight for self-hosting.

Product overview

Zentropi is a unified AI platform for building custom content labelers (AI guardrails) for text and image classification. The core offering consists of the Zentropi Platform with two interface modes: Web Build (no-code web interface) and Agent Build (coding agent integration via skill). Supporting the platform are two API products: Build API (create/test/benchmark labelers) and Run API (classify text or images against deployed labelers). The platform is powered by the proprietary CoPE-B model, a 4B parameter open-weight language model specialized for content labeling tasks.

Differentiator

Problem solved

Functional benefit

Products and services

• Zentropi Platform Unified AI platform for building, testing, and deploying custom content labelers (AI guardrails) for text and image classification, accessed via web, AI coding agent, or APIs.
• Build API REST API endpoint for programmatically creating, testing, and benchmarking custom content labelers, authenticated with Zentropi API keys (starting with zt_).
• Run API REST API endpoint for classifying text or image inputs against deployed content labelers in production environments, authenticated with Zentropi API keys (starting with zt_).
• CoPE-B Model Proprietary 4B active-parameter small language model specialized exclusively for content labeling, available as an open-weight model for free self-hosted deployment on customer infrastructure.

Quantifiable outcome

• 4B parameters - less than 1% the size of a standard LLM

Named customers5 records

Segments3 records

Ideal customer profiles4 records

Technology focussed Yes

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

Integration4 records

AI capability3 records

Feature6 records

Partnerships

Four partnerships are on record, tiered core.

• Claude CodecoreTechnology or IntegrationClaude Code is an AI coding agent by Anthropic. Zentropi provides a skill package that integrates with Claude Code, allowing users to instruct the agent to build, test, optimize, and deploy labelers using natural language prompts.
• CodexcoreTechnology or IntegrationCodex is an AI coding agent by OpenAI. Zentropi's skill package integrates with Codex, enabling developers to create content labelers through natural language instructions to the coding agent.
• HermescoreTechnology or IntegrationHermes is an AI coding agent. Zentropi skill integration allows Hermes to build, test, optimize, and deploy content labelers via natural language prompts from users.
• OpenClawcoreTechnology or IntegrationOpenClaw is an AI coding agent. Users can instruct OpenClaw to install the Zentropi skill and build custom labelers for specific use cases like flagging harassing comments.

Scale indicators1 record

Recent moves6 records

Expansion highlights6 records

Emerging players

• Guardrails AI: Open-source framework for adding validation and guardrails to LLM outputs. Comparable on the developer-first, open AI-guardrails approach, though focused on output validation rather than a specialized classifier model.
• Checkstep: Content moderation platform combining AI classifiers with policy workflows for online platforms. Comparable target market of platform/community operators needing customizable moderation.
• Cinder: Trust & safety operating platform for policy enforcement and content review. Overlaps on community moderation and policy enforcement use cases targeted by Zentropi.
• Modulate: AI startup focused on voice content moderation for online communities and games. Adjacent trust & safety classification player overlapping on community moderation, in a different modality.
• Unitary: AI startup providing multimodal (image/video) content moderation via contextual models. Comparable on purpose-built classification models for content safety at scale.

Direct peers

• Lakera: Builds AI guardrails to protect LLM and agent applications from unsafe content and prompt attacks. Directly comparable on the 'AI guardrails' positioning Zentropi emphasizes for AI systems.
• WebPurify: Provides text and image moderation APIs for online platforms. Directly comparable as an API-delivered content classification/moderation service.
• Hive AI: Provides cloud APIs for automated content moderation of text, image, and video. Directly comparable as an AI-model-based classification service for trust & safety, the same core problem Zentropi addresses.
• Sightengine: Offers API-first image, video, and text moderation with customizable rules. Closely mirrors Zentropi's developer-focused, API-delivered content classification model.

Broad incumbents

• ActiveFence: Established trust & safety and content moderation platform (including the former Spectrum Labs) serving large platforms. Overlaps directly on AI-driven content classification but as a broader, more enterprise-entrenched suite.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat5 records

Key risks6 records

Key highlights7 records

Customer concentration

• GitHubgithub.com/zentropi-ai/skills
• LinkedInlinkedin.com/company/zentropi

Revenue estimate

Valuation estimate

Number of profiles

Profiles1 record

Funding overview

Funding rounds

Investors

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Zentropi