LESSON 01
What is an agent?
You've probably used ChatGPT. You type something, it replies. That's a chatbot: one round-trip. You ask, it answers.
An agent is different. Instead of answering a question, it does a job.
Say you type: "Prep me for my 3pm meeting with Sarah at Northwind Foods." A chatbot would ask you to paste in everything you know about Sarah and Northwind. An agent already has your calendar, your CRM, your past meeting notes. It pulls the relevant pieces, reads through them, writes a brief, and hands it to you ready to use.
The difference comes down to three things:
Memory
An agent knows your context. Your contacts, your company, your voice, your recurring jobs. You load it once; it uses it every time.
Tools
An agent reads files, checks your calendar, queries a database, and writes output to a document. It acts on your tools, not just your chat window.
Instructions
An agent runs on a brief that tells it how to behave: what to prioritize, what format to use, what to ask if it's missing something. With the right brief, it does the actual job. Skip it and you get generic output.
Most people who install Claude or ChatGPT are using it as a chatbot. Question in, answer out. That's useful, but it's a fraction of what's possible.
Getting an agent to actually perform requires setting it up right: the instructions, the memory, how the job is structured. tigertiger builds and configures that so you don't start from scratch.