Getting Started

Core Concepts

This page explains the core building blocks of Sprig and how they work together. Understanding these concepts will make the rest of the docs much easier to follow.

Agents

An agent is an AI worker with a clear responsibility.

Agents don’t just respond — they operate within boundaries you define.

What an Agent Includes

Each agent has:

Instructions – What the agent should do
Behavior Controls – Tone, creativity, limits
Memory – What it can remember between runs
Knowledge Access – Data the agent can reference
Tools – Actions the agent can take

{
  "name": "Support Agent",
  "instructions": "Answer customer questions using only approved documentation.",
  "temperature": 0.2,
  "memory": "session"
}

UI vs API Usage

Sprig supports both visual and programmatic workflows.

UI-first

Faster setup
Good for iteration
No code required

API-first

Full control
Better for production
Easier to automate

Example API usage:

POST /v1/workflows/run
Authorization: Bearer YOUR_API_KEY

Common Mistakes to Avoid

Overloading one agent with too many roles
Putting long instructions instead of structured knowledge
Skipping workflows and hardcoding logic
Using memory where it’s not needed

Quick Start

What Is an Agent?