Building your first agent
This cookbook provides a comprehensive guide to creating, evaluating, and deploying your first Chicory AI agent. You'll learn the complete Agent Development Life Cycle (ADLC) from initial creation to production deployment.
Whether you're new to AI agents or transitioning from other platforms, this step-by-step guide will help you build a robust, production-ready agent.
Quick Start
Log into the Chicory AI dashboard and set up your organization/project
Create your first agent with proper integrations (data, code, documents, tools)
Define evaluation criteria and upload a validation dataset
Run evaluations to test and improve your agent iteratively
Deploy your agent using the REST API or MCP Gateway with proper authentication
Contents
Agent Creation - Setting up your first Chicory agent with integrations
Evaluation - Testing and iterating on your agent using validation datasets
Deployment - Deploying your agent to production using the REST API or MCP Gateway
What You'll Learn
Agent Creation: How to configure integrations, create agents, and add optional MCP tools
Evaluation Framework: Understanding the Agent Development Life Cycle (ADLC) and validation processes
Iterative Improvement: Using evaluation results to evolve your agent through multiple iterations
Production Deployment: Securely deploying agents using API tokens and proper authentication
Prerequisites
Access to Chicory AI dashboard
Basic understanding of AI agents and their use cases
API token for deployment (generated from Chicory platform)
Agent Development Life Cycle (ADLC)
This cookbook follows the complete ADLC process:
Build - Create your agent with proper configurations
Evaluate - Test against validation datasets with defined criteria
Evolve - Iterate based on evaluation results and feedback
Deploy - Push to production using REST API or MCP Gateway
Monitor - Track performance and make ongoing improvements
Last updated