Agents Overview
The Agents module is the central control panel for the SpendCraft AI. Agents are machine learning models designed to understand your specific business context, learning from your corrections to automate the classification of millions of spend transactions with high precision.
🤖 How Agents Work
Agents act as the bridge between your raw data and your structured taxonomy. They analyze transaction descriptions, vendor names, and GL codes to predict the correct category.
🔗 Quick Links
Concepts & Structure
Understand the relationship between Agents, Training Datasets, and Taxonomies.
Create a New Agent
Initialize your AI model and define the parameters for its initial learning phase.
Audit & Configure
Refine Agent accuracy by reviewing low-confidence predictions and providing feedback.
Publish to Workspace
Deploy your classified data to Workspaces for executive reporting and team collaboration.
💡 The Feedback Loop
The power of a SpendCraft Agent lies in Continuous Learning.
- Predict: The Agent classifies a dataset.
- Audit: You correct any misclassifications (e.g., changing "Software" to "Hardware").
- Learn: The Agent updates its logic based on your correction, ensuring it never makes the same mistake twice.
Next Step: New to AI classification? Start with Agent Concepts & Structure.