Skip to main content

Glossary

Quick reference for AgentFlow terminology.

A

Agent
A configured AI assistant with specific behavior, personality, and capabilities.
API
Application Programming Interface - allows external applications to interact with AgentFlow programmatically.

C

Context Window
The amount of conversation history an AI model can remember at once.
Conversation
A thread of messages between a user and an agent.

D

Distribution
Deploying an agent to multiple organizations while maintaining data isolation.

M

Model The underlying AI technology (e.g., GPT-4, Claude) that powers an agent. Model Configuration A complete set of settings for an AI model including endpoint, headers, request schema, and message formatting rules. Model Export Downloading a model configuration as a YAML file for backup or sharing. Header values and API keys are excluded for security. Model Import Uploading a YAML configuration file to create or update a model. Requires manually adding API keys and header values after import.

O

Organization
A workspace containing agents, team members, and conversations with complete data isolation.

R

Role
Permission level assigned to team members (Owner, Admin, Member, Guest).

S

System Prompt
Instructions that define an agent’s behavior, personality, and capabilities.

T

Temperature Setting that controls how creative vs. consistent an agent’s responses are (0.0-1.0). Token Unit of text processed by AI models. Roughly 0.75 words per token.

Y

YAML A human-readable data format used for model configuration files. Allows easy sharing and version control of model settings.

Need More Help?

Contact support or visit our documentation
I