Emcie Documentation
Emcie is an auto-optimizing inference platform for Parlant agents. It automatically optimizes your agent's inference through dynamic prompt tuning and SLM distillation, delivering LLM-level accuracy at significantly reduced costs.
Contents
- Getting Started - Set up Emcie with your Parlant agent
- Pricing - Understand the cost structure
- Rate Limits - API limits and usage tiers
- Model Tiers - Choose between Jackal and Bison
- Model Roles - Learn about Teacher/Student optimization
Getting Started
Get up and running with Emcie in minutes. Set your API key, configure Parlant to use Emcie's NLP service, and deploy your agent.
Install, configure, and deploy your first Emcie-powered agent.
What you need before getting started.
Pricing
Emcie uses token-based pricing with significant savings as your agent transitions from Teacher to Student models.
Understand costs for Jackal and Bison tiers.
How Emcie reduces your costs over time.
Rate Limits
API rate limits protect the platform and ensure fair access for all users. Limits increase as your usage grows.
Model Tiers
Choose the right balance between cost and accuracy for your use case.
Lower tier optimized for cost efficiency. Models from 0.3B to 20B parameters.
Higher tier optimized for accuracy. Models from 20B to 180B parameters.
Model Roles
Understand how Emcie's Teacher/Student architecture automatically optimizes your inference costs.
How distillation and prompt optimization work.
The data collection phase before optimization.
How Emcie handles updates to your agent.
How your data is protected.