AI Agents Connected to Your Tools

Course outline

An introduction lesson, then four modules: 24 lessons in all. Open a module to see its lessons.

1. 0Module 0Introduction lesson1 lesson · 8 min

   1. 0.1Read, write on your side, send Free sample lessonwhat a connected agent changes, and what this course teaches you8 min
2. 1Module 1What Claude and ChatGPT do on their own, then n8n: open your accounts, build a first complete workflow6 lessons · 48 min

   1. 1.1Before n8nwhat Claude and ChatGPT already do on their own12 min
   2. 1.2n8n Cloud, what it costs, and the accounts to opena Google account for the course, OpenAI with a spending limit9 min
   3. 1.3Anatomy of a workflowtrigger, steps, action, and what travels between them5 min
   4. 1.4The trigger and the AI stepread the news, have it summarized by OpenAI9 min
   5. 1.5The actionwrite a row in a spreadsheet, and the connection to Google7 min
   6. 1.6Test, read the log, put into service; what one execution costs6 min
3. 2Module 2Three agents connected to your tools9 lessons · 1 h 09

   1. 2.1Agent 1, email and calendarthe mission, the kit and the diagram6 min
   2. 2.2Agent 1, the buildsort the emails, read the calendar, prepare the draft that offers three time slots10 min
   3. 2.3Agent 1test on fifteen emails, correct, put into service6 min
   4. 2.4Agent 2, spreadsheetthe mission, the sheet and the diagram6 min
   5. 2.5Agent 2, the buildenrich row by row without overwriting the source, summarize the week, alert12 min
   6. 2.6Agent 2test on the fictional sheet, correct, schedule7 min
   7. 2.7Agent 3, website formthe mission, the form and the diagram6 min
   8. 2.8Agent 3, the buildreceive the form, filter out bots, qualify, prepare the reply, create the client record10 min
   9. 2.9Agent 3test on twelve requests, correct, put into service; the shop variant7 min
4. 3Module 3Memory, decisions and handing over to a person5 lessons · 29 min

   1. 3.1Memorywhat the agent remembers, where, and for how long5 min
   2. 3.2Decideconditions, branches, who decides what, and when to let the model choose its tools8 min
   3. 3.3Approve before sendingthe approval request5 min
   4. 3.4When the agent hesitatesthe queue4 min
   5. 3.5What goes wherepersonal data, AI providers’ terms, API keys7 min
5. 4Module 4Errors, costs and monitoring3 lessons · 18 min

   1. 4.1When a workflow failsretry, error workflow, alert5 min
   2. 4.2What an agent coststhe calculation before you build, the limits, the runaway loops6 min
   3. 4.3Monitor, update, and the “before you let an agent run” checklist7 min

Objectives

By the end of this course, you will be able to:

• Say what Claude and ChatGPT do on their own with Gmail and a scheduled task, what they cannot do, and when an automation tool becomes necessary
• Build a complete workflow in n8n (trigger, AI step, action), test it step by step, put it into service and read its log
• Connect an agent to your email and calendar to sort messages, prepare replies and suggest time slots, without sending anything on its own
• Have an agent read, enrich and summarize a spreadsheet, and raise an alert based on rules written in advance
• Receive the requests from your website form, qualify them, prepare the reply and keep the client record
• Decide what the workflow, the AI model or a person should handle, and give an agent a memory when it helps
• Add a human approval step before anything goes to a third party, and a queue for the cases where the agent is unsure
• Handle errors, cap AI spending and monitor a live agent
• Say what data your agent sends, to whom and in which country, and keep your API keys safe
• Estimate, before you build, what an agent will cost each month, and adapt the kits provided to your own work