This lesson is being recorded
Every video of this course is being uploaded to my YouTube channel. Subscribe so you do not miss it.
Bruno Ashford @brunoashfordSubscribeModule 1 · Lesson 1.1 · 10 min
From Chatbot to Agent
Understand what changes when AI stops answering questions and starts doing work.
Use ← → to move. The marker highlights text and the pencil draws on the slide; navigation locks while you annotate. Undo removes the last mark, Esc exits.
1Know the loop
Every agent works in a loop: it reads your request, gathers context (files, data), takes an action, checks the result, and decides the next step. It keeps going until the job is done or it needs you. A chatbot does one step: it replies.
2Notice the three things an agent has that chat does not
Tools (it can read and write files, run commands, search), memory of your business (instructions it reads every session), and the ability to work across many steps without you copying anything. Those three are why the output is finished work, not a draft you still have to process.
3Change how you ask
With chat you ask for content. With an agent you ask for an outcome and describe the finish line. Compare these two requests for the same job.
Chat style: Write a recall reminder for a dental patient.
Agent style: In the file patients-due-october.csv, find every patient due a check-up this month. Draft a personalised recall reminder for each one using our template in reminder-template.md, and save them all in one file called october-reminders.md. Tell me how many you drafted and list any rows with missing data.4Expect a report, not just an answer
A good agent tells you what it did. Get used to asking for a short report at the end: files read, files created, what it skipped and why. That report is how you manage it.
5Remember who is accountable
The agent does the steps. You still own the result. An agent is closer to a capable new hire than to software: fast, tireless, occasionally wrong, and much better with a clear brief.
A dental clinic owner told me she had been using AI for a year and it had saved her almost nothing. I asked her to show me. She would ask a chatbot to write a recall reminder, copy it, open the practice software, paste it, fix the patient name, and repeat. Forty times. The AI was helping. She was still doing the work.
An agent closes that gap. Instead of giving you text to carry around, it has access to tools: it can open files, read a spreadsheet, run a step, check the result and try again. You describe the outcome, it works through the steps, and it comes back with the finished thing. That loop of plan, act, check and repeat is what the word agentic means.
Key points
- Chat replies once. An agent plans, acts, checks and repeats.
- Agents use tools on your files instead of handing you text.
- Ask for the outcome and describe the finish line.
- Always ask for a short report of what was done.
- You stay accountable for the result.
Avoid
- Using an agent like a chatbot, one small question at a time.
- Describing the task instead of the finished result.
- Accepting the output without asking what it did and what it skipped.
The clinic moved its monthly recall process to an agent. The receptionist exports the list of patients due, the agent drafts 40 personalised reminders in the clinic's tone, flags three patients with missing phone numbers, and saves everything in one file for review. What used to be most of a Monday morning is now a 15-minute review. The receptionist still sends them, but she no longer writes them.