Getting Started

Start using ClassifAI in minutes - classify text and images with automatic label inference

What is ClassifAI?

ClassifAI is a self-improving classification API for text and images. Use it to categorize content, route requests, analyze sentiment, moderate content, and more - without training your own models.

Key Features:

Classify text and images (separately or together)
Automatic label inference from descriptions
Self-improving through feedback loops
Anonymous and authenticated tiers
Model Context Protocol (MCP) support for AI agents

Installation

No installation needed. Just use curl or any HTTP client.

Install the Python SDK:

pip install classifai-sdk

Quick Start

1. Make Your First Classification

curl -X POST https://api.classifai.dev/classify \
  -H "Content-Type: application/json" \
  -d '{
    "content": [
      {"type": "text", "content": "This product is amazing! Fast shipping too."}
    ],
    "labels": ["positive", "negative", "neutral"]
  }'

from classifai import ClassifAI

client = ClassifAI()

result = client.classify(
    content="This product is amazing! Fast shipping too.",
    labels=["positive", "negative", "neutral"]
)

print(f"Label: {result['label']}")
print(f"Confidence: {result['labels'][result['label']]:.0%}")

Response:

{
  "labels": {
    "positive": 0.92,
    "negative": 0.05,
    "neutral": 0.03
  },
  "label": "positive",
  "detection_id": "det_abc123...",
  "project_id": "proj_xyz789",
  "labels_used": ["positive", "negative", "neutral"],
  "ground_truth_url": "https://api.classifai.dev/ground_truth/det_abc123...",
  "processing_time_ms": 1166
}

2. Provide Feedback to Improve

Help the model learn by providing the correct label:

curl -X POST https://api.classifai.dev/ground_truth/det_abc123... \
  -H "Content-Type: application/json" \
  -d '{"ground_truth": "positive"}'

feedback = client.submit_feedback(
    detection_id=result["detection_id"],
    ground_truth="positive"
)

The system learns from this feedback and improves accuracy for your specific use case.

3. Use Automatic Label Inference

Instead of providing labels, describe what you want to classify:

curl -X POST https://api.classifai.dev/classify \
  -H "Content-Type: application/json" \
  -d '{
    "content": [
      {"type": "text", "content": "The food was cold and service was slow"}
    ],
    "description": "Restaurant reviews with quality indicators"
  }'

result = client.classify(
    content="The food was cold and service was slow",
    description="Restaurant reviews with quality indicators"
)

print(f"Inferred labels: {result['labels_used']}")

The system will automatically infer appropriate labels like "poor_quality", "slow_service", etc.

Access Tiers

Tier	API Key	Rate Limits	Request Size Limit	Cost
Free (Test)	Optional	10/min, 100/day	500 KB	Free
Hobby	Yes	10/min, 1,000/day	2.5 MB	$5/month
Production	Yes	100/min, 10,000/day	10 MB	$25/month

Note: Request size limits apply to the entire request body, including all text and base64-encoded images. For tips on reducing request sizes, see Image Optimization.

Use Cases

Sentiment Analysis

Classify customer reviews, support tickets, or social media posts.

Content Moderation

Automatically flag inappropriate content, spam, or policy violations.

Request Routing

Route customer inquiries to the right team based on intent and urgency.

Image Classification

Classify product images, detect content types, or categorize visual content.

Combine text and images for richer classification (e.g., product reviews with photos).

Next Steps

View API Reference - Complete endpoint documentation
See Examples - Real-world use cases and code samples
MCP Integration - Use ClassifAI with AI agents like Claude

Get an API Key

Ready to try it out? Sign up now to get started for free.

On this page