Read the narration transcript
Welcome to lesson six, Your first workflow. This last lesson pulls everything together. You'll turn one recurring task into a repeatable AI-assisted routine, and you'll get a plain introduction to automation and AI agents, two terms you're likely to hear at work.
Asking AI for help now and then is useful. The bigger gain comes when you turn a task you do every week into a routine. The same steps, the same saved request, the same check, every time. That makes the results more consistent and the time savings easier to see. Start with the task you picked in lesson five, or any recurring, low-risk, text-heavy task.
There are four steps. First, write it down. List how you do the task today, step by step. Note where the information comes from, what you produce, who receives it, and how long it takes. Second, build a saved prompt or template. Write a full request with context, role, task, format and constraints, just like in lesson two, and leave clearly marked blanks for the parts that change, like this week's notes or the date range. Third, add the check step. Using what you learned in lesson three, decide in advance what you'll verify each time, which numbers, names or dates must be confirmed, and against what source. Write it into the routine so it isn't optional. Fourth, time it. Run the routine a few times and compare the time against your old way, including the review. If it isn't saving time, or the quality drops, adjust the prompt or pick a different task.
Once a routine is stable, parts of it may be automated. Workflow automation tools, such as Power Automate or Zapier, connect the software you already use and move information between them based on rules. For example, when a form is submitted, save the response and send a notification. Some of these tools can now include an AI step, such as summarizing the form response before it's saved. Automation is usually set up by someone with the right access and approval, often IT or a designated team member. Your job is to know what's possible, and to bring good, well-documented routines to the people who can automate them.
You'll also hear about AI agents. An agent is an AI system that can carry out a multi-step task with some independence. That independence is exactly why human approval matters. A well-designed workflow puts a person at the steps that matter. Before anything is sent outside the organization. Before money moves. Before records change. Before a decision affects someone. The same rules from this course apply. You own the result, sensitive data stays protected, and the check step is never skipped.
Below this section, an interactive follows one task, a monthly vendor spend summary, through four levels of help. Pause the audio here and step through it. At level one, you paste last month's vendor spend into a one-off prompt, and every time starts from scratch. At level two, you reuse a saved template and a checklist each month. That's faster and more consistent, but still manual. At level three, it's an automated workflow. An automation tool pulls the export and sends it to the AI on the first business day. You review and approve before it goes to anyone. At level four, an agent gathers the data, drafts the summary, flags anything unusual and prepares the email. You set its limits, and you approve the flagged items and the final send. Most teams get real value from levels one and two long before they need automation or agents.
The worked example is a saved prompt for a weekly status note to your manager. Using only your notes, it asks for a one-sentence overall status, up to four bullets on what was completed, up to three on what's next, and any risks or blockers, all in under one hundred seventy-five words. Its blanks make it reusable every week. The check step is built in. Confirm every completed item and date against your actual notes before sending. Now the final Try it exercise, Build and run your routine. Choose one recurring, low-risk task, and write down your current steps and timing. Write a saved prompt with marked blanks, and a short checklist of two or three things you'll verify every time. Run it once with non-sensitive information in your approved assistant, check included. Note the time and one change you'd make. If it was faster with the same or better quality, use it again next week. If not, adjust the prompt or try a different task.
Here's what to take with you. Write the task down, save a prompt, build in the check, and time it. Automation and agents extend good routines. They don't replace review. And keep a person approving the steps that matter. Take the last quiz below.
That's the end of AI Basics. Once you've passed all six lesson quizzes, you can add your name and print or save your certificate of completion from BoulevardLabs. If you'd like to keep practicing, the free AI Fluency Lab on our website has guided exercises. And if your team wants to build these skills together, BoulevardLabs also offers instructor-led training. The habits from this course carry over. Ask well. Check the work. Protect the data. And keep a person in charge of what matters. Thank you for taking the course.
From one-off help to a routine
Asking AI for help now and then is useful. The bigger gain comes when you turn a task you do every week into a routine: the same steps, the same saved request, the same check, every time. That makes the results more consistent and the time savings easier to see.
Pick the task you identified in Lesson 5, or any recurring, low-risk, text-heavy task.
Four steps to build it
- Write it down. List how you do the task today, step by step: where the information comes from, what you produce and who receives it. Note how long it usually takes.
- Build a saved prompt or template. Write a full request with context, role, task, format and constraints, and leave clearly marked blanks for the parts that change, such as [paste this week's notes] or [date range]. Store it somewhere you can find it.
- Add the check step. Decide in advance what you will verify each time, based on Lesson 3: which numbers, names or dates must be confirmed and against what source. Write it into the routine so it is not optional.
- Time it. Run the routine a few times and compare the time against your old way, including the review. If it is not saving time or the quality drops, adjust the prompt or pick a different task.
What comes next: automation
Once a routine is stable, parts of it may be automated. Workflow automation tools, such as Power Automate or Zapier, connect the software you already use and move information between them based on rules. For example: when a form is submitted, save the response to a spreadsheet and send a notification. Some of these tools can now include an AI step, such as summarizing the form response before it is saved.
Automation is usually set up by someone with the right access and approval, often IT or a designated team member. Your job as a beginner is to know what is possible and to bring good, well-documented routines to the people who can automate them.
What comes next: AI agents
You will also hear about AI agents. An agent is an AI system that can carry out a multi-step task with some independence: look something up, draft a document, update a record, then report back. Instead of answering one request, it works through a sequence.
That independence is why human approval matters. A well-designed workflow puts a person at the steps that matter: before anything is sent outside the organization, before money moves, before records change, before a decision affects someone. The same rules from this course apply. You own the result, sensitive data stays protected and the check step is never skipped.
Loading the interactive example…
Where to go from here
To keep practicing, visit the AI Fluency Lab for guided exercises. If your team wants to build these skills together, BoulevardLabs offers instructor-led training. Either way, the habits from this course will carry over: ask well, check the work, protect the data and keep a person in charge of what matters.
A saved prompt for a weekly status note
The blanks make it reusable every week. The fixed format makes notes easy to compare over time. The check step is built in: confirm every completed item and date against your actual notes before sending.
Build and run your routine
- Choose one recurring, low-risk task and write down your current steps and how long it usually takes.
- Write a saved prompt for it with marked blanks for the parts that change.
- Write a short checklist of two or three things you will verify every time.
- Run the routine once with non-sensitive information in your approved assistant, including the check.
- Note the time it took, and one change you would make to the prompt for next time.
How you’ll know it worked: You should have three saved things: the prompt, the checklist and a time comparison. If the routine was faster with the same or better quality, use it again next week. If not, adjust the prompt or try a different task.
Use only non-sensitive information, in a tool your organization allows.
Key takeaways
- Write the task down, save a prompt, build in the check and time it.
- Automation and agents extend good routines; they do not replace review.
- Keep a person approving the steps that matter.





