Foundations Module 5 · Cheat Sheets (Reference)

Monday Morning Wins

Last reviewed

Beginner

What you'll learn

~15 min
  • Run five workspace-agent workflows on your own files and get a visible result
  • Use plan mode to let the agent ask its questions before it builds
  • Check the output file before you share it
  • Frame your results in a way that demonstrates value to your team

These five workflows produce visible results you can share with your team this week. Each takes under 30 minutes and uses the workspace agent you already installed (a desktop app or a CLI), pointed at a real file from your normal work.

💬This is about proving value, not just learning

The skills in this course only matter if you use them. Pick the workflow that matches your actual work, do it today, then tell someone what you did.

The pattern behind all five

Every win is the same short loop, the build loop at small scale:

  1. Put the real file in a folder. Make a new folder, such as monday-wins, and drop in the file the agent should work on.
  2. Open your agent in that folder and switch to Plan mode.
  3. Send one or two plain sentences saying what you want and what file to write.
  4. Answer its questions. In plan mode it asks what it needs, so you don’t have to guess at details.
  5. Check the output file it wrote, against your original, before anyone else sees it.
  6. Keep what works. Ask the agent to save a checkpoint, and tell it what to change if something is off.
⚠Privacy check

Before putting any file in the folder, check it for confidential information: employee names in sensitive contexts, financial details, trade secrets, or anything your organization treats as restricted. When in doubt, remove identifying details or ask your IT or compliance team which AI tools are approved.


1. Automate your weekly status report

Put in the folder: notes.txt, your raw notes, calendar highlights or bullet points from the week.

Send:

Turn notes.txt into a weekly status report in status.md with the sections Completed, In Progress, Blocked and Next Week. Keep it to short bullets.

It may ask: who the audience is, or whether to keep names. Answer in a few words.

Check status.md: Are the facts accurate (rephrasing can change meaning)? Did anything you forgot make it in, or something real get left out? Is the tone right for your audience?

Share it as your normal weekly update. If someone comments on how organized it is, that’s your opening to mention the workflow.

🧬In Your Field: Biotechclick to expand

Lab version: use your lab notebook entries as notes.txt and ask for the sections Experiments Run, Results, Issues, Next Steps, with sample IDs where relevant.

📊In Your Field: MIS / Businessclick to expand

Business version: include KPI numbers and milestones, and ask for a “Key Metrics” section at the top with percentage changes from last week.


2. Turn messy data into a chart

Put in the folder: sales.csv, an export from your own system, however messy.

Send:

Look at sales.csv and make chart.html: a clean chart of sales over time, plus two sentences on the main trend.

It may ask: which columns matter, how to treat blank or odd rows, what kind of chart you prefer.

Check chart.html: Open it in your browser. Do a few bars or points match numbers you can find in the original rows? Are the axes labelled? Did it quietly drop rows? Ask “How many rows did you skip, and why?”

Share it as a screenshot with the one-line trend. Attach the CSV so others can check.


3. Summarize a long document

Put in the folder: the report, transcript or meeting notes (as report.pdf, report.docx or a text file).

Send:

Read the document in this folder and write summary.md: a five-bullet summary, the decisions made, and action items with owners.

It may ask: who the summary is for, and how long it should be.

Check summary.md: Does it capture the main points? Are the action items attributed to the right people? Did it flag real ambiguities rather than guess?

Share it as “meeting summary” or “document highlights” with the original linked so people can check details.


4. Draft a professional email from bullet points

Put in the folder: bullets.txt, your rough points, plus who the email is for.

Send:

Turn bullets.txt into a short, friendly, professional email in email-draft.md. It is going to my manager.

It may ask: what you want the reader to do, and by when.

Check email-draft.md: Is every point there, and is nothing added that you didn’t say? Is the ask clear, and does the tone sound like you?

Share it by pasting it into your email client and editing the last line yourself. You send it, not the agent.


5. Turn a vague request into a project brief

Put in the folder: request.txt, the vague ask you were handed, word for word.

Send:

Read request.txt and write brief.md: the problem, the scope, how we will measure success, and the first three steps.

It may ask: the deadline, the audience and what is out of scope. This is the useful part: its questions are the ones you’d have to ask anyway.

Check brief.md: Does the problem statement match the real issue? Is the scope realistic (agents are often too ambitious; trim it)? Are the success measures measurable, and are the first steps things you can do this week?

Share it with the person who made the request: “Here’s how I’m thinking about this. Does this match what you had in mind?” That turns a vague task into a productive conversation.

🏛️In Your Field: Government Devclick to expand

Government version: ask for a “Compliance considerations” section: which systems are affected, whether a security review is needed, and any procurement implications.


Make your own

The pattern works for any task where you turn one file into another:

  1. Pick a task you repeat that turns information from one format into another.
  2. Run it once in plan mode and note what you had to correct.
  3. Keep the refined instruction. Ask the agent to save it as a skill or note in the project (Giving Good Instructions explains why a reused prompt becomes a skill).

Five workflows saving 30 minutes each adds up to over 10 hours a month, a full workday back.

Key takeaways

  • Start with real work. Use an actual file from your week, not a toy example.
  • Plan mode first. Two plain sentences and the agent’s questions beat a long prompt.
  • Check the file it wrote before sharing. Agent output is a draft, not a final product.
  • Show results, not tools. Lead with the outcome (“I cut our weekly report to 15 minutes”, with your own measured number once you have one).
  • Keep what works as a checkpoint or a saved skill, then do one win today and tell someone.
KNOWLEDGE CHECK

What's the most effective way to share an AI-assisted win with your team?

Search lessons