Welcome

Introduction:

Platform

FAQ

Plans and Pricing

Changelog

Agents:

Agents Overview

Build Your Agent

Share Your Agent

Monitor Your Agent

Features:

Interaction Tools (100+)

Voice & Arabic support

Environment Awareness

System Prompting

Templates (20+)

Available LLMs

Shipcrew

Tips & Tricks:

Learn Agent

Prompting Guide

Slack Integration

Alternatives

Use Cases

TestAI Integration

Referral Program

About:

Team

Support Policy

Privacy Policy

Terms & Conditions

CNTXT

Basics of Agent

This page introduces the essential building blocks of creating AI agents in Shipable. If you're new here, this is your cheat sheet.

What is an Agent?

A Shipable agent is an intelligent, interactive assistant that can:

• Understand and respond to user input
• Access knowledge from documents or databases
• Trigger workflows and tools (like Slack, Notion, Stripe)
• Maintain memory across interactions
• Operate across web, Slack, WhatsApp, API, voice, and more

In AI terms, an agent is an autonomous entity that perceives, reasons, and acts in an environment to achieve specific goals.

Core Concepts

1. Prompt (Instructions)

This is the agent’s brain. Prompts define its role, behavior, tone, and rules. Good prompts = good agents.

Think of it as the agent’s job description.

2. Knowledge Base (KB)

Attach data to your agent so it can answer questions or make decisions. This can include:

• PDFs, Docs, Notion pages
• Q&A pairs
• Website content
• Google Drive or Airtable integrations

Agents use this as a structured or unstructured repository for reasoning and decision-making.

3. Tools

Give your agent real-world powers. With tools, your agent can:

• Send emails (Gmail)
• Look up calendar availability (Cal.com)
• Create tasks (Jira, Asana)
• Fetch data (APIs, CRMs)
• Trigger payments (Stripe)

Tools are how agents act in the environment after reasoning.

4. Memory

Memory lets your agent retain information about users and interactions across sessions. Great for:

• Personalization
• Long workflows
• Ongoing chats

Memory enables continuity, learning, and richer state-based interactions.

How It All Comes Together

Here’s what happens under the hood:

1. Perception: Agent receives input from the environment.
2. Prompt Logic + State: Interprets it using its instructions and current state.
3. Memory + KB: Pulls context from memory and external data sources.
4. Planning/CoT/ReACT: Applies logic to figure out next steps, possibly breaking it down (Chain of Thought) or combining thought and action (ReACT).
5. Tools + Action: Executes real-world actions.
6. Logging + Learning: Updates memory, evaluates outcome, and (optionally) learns.