Big Model Management Revolution: RagaAI Catalyst Improves AI Efficiency by 300%

• •  Project Management  - End-to-end project lifecycle management
• •  Dataset management  - multi-format dataset support with automatic schema mapping
• •  Evaluation management-  multi-dimensional model evaluation indicator system
• •  Tracking management-  full-link call tracking and analysis
• •  Prompt management  - versioned prompt templates and dynamic compilation
• •  Synthetic data  - Intelligent question-answer pair generation and data augmentation
• •  Safety guardrails  - multi-layered protection strategy and real-time execution
• •  Red Team Testing  - adversarial testing and vulnerability scanning

Install

Install using pip:

pip install ragaai-catalyst

Configuration

Configure authentication credentials:

from  ragaai_catalyst  import  RagaAICatalyst

# Initialize the client
catalyst = RagaAICatalyst(
    access_key= "your access key" ,
    secret_key= "your security key" ,
    base_url= "API endpoint"
)

Steps to obtain the key :

1. 1. Log in to the RagaAI console
2. 2. Go to "Personal Settings" → "Authentication Management"
3. 3. Click Generate New Key

Note: All API operations require authentication credentials

Core Features

project management

# Create a new project
project = catalyst.create_project(
    project_name = "Intelligent Customer Service System" ,
    usecase= "chatbot"
)

# List all projects
projects = catalyst.list_projects()

Dataset Management

Support CSV/JSONL/DataFrame multiple data formats:

from  ragaai_catalyst  import  Dataset

ds = Dataset(project_name= "Intelligent Customer Service System" )

# Create a dataset from CSV
ds.create_from_csv(
    csv_path= "conversation record.csv" ,
    dataset_name = "Customer Service Dialogue" ,
    schema_mapping={ 'User question' :  'query' ,  'Robot reply' :  'response' }
)

Assessment Management

from  ragaai_catalyst  import  Evaluation

eval  = Evaluation(
    project_name = "Intelligent Customer Service System" ,
    dataset_name = "Customer Service Dialogue"
)

# Add evaluation metrics
eval .add_metrics([
    {
        "name" :  "Factual Accuracy" ,
        "config" : { "model" :  "gpt-4o" ,  "threshold" : { "gte" :  0.8 }}
    }
])

# Get the evaluation results
results =  eval .get_results()

Tracking management

from  ragaai_catalyst  import  Tracer

tracer = Tracer(
    project_name = "Intelligent Customer Service System" ,
    dataset_name = "Service Tracking"
)

with  tracer():
    # Business logic that needs to be tracked
    response = chatbot.query( "How do I reset my password?" )

Agent Tracking

@trace_Agent( name= "Recommended Agent" )
class RecommendationAgent : 
    def recommend ( self, text ): 
        # Business Logic
        current_span().add_metrics(accuracy= 0.92 )

Prompt Management

from  ragaai_catalyst  import  PromptManager

pm = PromptManager(project_name= "Intelligent Customer Service System" )
prompt = pm.get_prompt( "Standard reply template" )

# Dynamic compilation tips
compiled_prompt = prompt. compile (
    query = "Order Query" ,
    context = "User needs to view recent orders"
)

Synthetic Data Generation

from  ragaai_catalyst  import  SyntheticDataGeneration

sdg = SyntheticDataGeneration()
text = sdg.process_document( "Product Manual.pdf" )

# Generate complex question-answer pairs
qna_data = sdg.generate_qna(text, question_type= 'complex' , n= 50 )

Guardrail Management

from  ragaai_catalyst  import  GuardrailsManager

gm = GuardrailsManager(project_name= "Intelligent Customer Service System" )

# Add security rules
gm.add_guardrails(
    deployment_id = 123 ,
    guardrails=[{
        "name" :  "Sensitive Information Filtering" ,
        "config" : { "threshold" : { "lte" :  0.1 }}
    }]
)

Red Team Testing

from  ragaai_catalyst  import  RedTeaming

rt = RedTeaming(model_name= "gpt-4" , provider= "openai" )

# Run a security scan
test_report = rt.run(
    description = "Recruitment Consultant Robot" ,
    detectors=[ "Bias Detection" ,  "Harmful Content" ],
    response_model=chatbot.predict
)