Start with Recurring Tasks

Why start with recurring tasks

Professionals who get the most value from ChatGPT rarely start with an ambitious project. They start with tasks they already do, over and over, every week. This is also the progression OpenAI Academy proposes in its courses for the workplace, launched in June 2026: first improve a daily task, such as drafting, summarizing, or preparing for a meeting; then turn the prompts that work into repeatable workflows; and finally lead more structured work with agents. This course follows the same logic, built on three habits: start with recurring tasks, save what works, and build value over time.

The three criteria for an ideal starter task

The first criterion is recurrence. The task comes up every week, or even every day: preparing a team check-in, drafting a follow-up email, sorting incoming requests. This repetition is your ally, because each iteration lets you refine your prompt and therefore improve the result. A task you never do again teaches you nothing.

The second criterion is a clear, defined outcome. The deliverable must be concrete: an email, a summary, a spreadsheet, an agenda. If you cannot describe the shape of the expected result, ChatGPT cannot either.

The third criterion depends on you, not the tool: you must have enough expertise on the subject to judge whether the output is correct, relevant, and usable. You remain the quality filter. Delegating a task you do not master is like publishing text no one has proofread.

What to delegate first

Certain tasks recur across professions. Meeting preparation: scattered notes become a structured agenda with decision points. Operational updates: raw data becomes a readable weekly briefing. Sorting incoming requests—emails, tickets, or queries—which ChatGPT classifies and prioritizes according to your criteria. Meeting minutes and follow-up emails complete this foundation.

These tasks offer an excellent effort-to-value ratio: ChatGPT produces a first draft quickly, and the result gains consistency from week to week because the structure does not change. The actual time saved depends on your role; Lesson 15 teaches you how to measure it rather than assume it.

What makes a start fail

The first reflex to correct is misplaced ambition: rewriting your entire marketing strategy with ChatGPT is a project, not a starting point; drafting your weekly follow-up email is a starting point. The second trap is choosing a one-off task, which deprives you of any chance to iterate. The third is neglecting context: ChatGPT does not guess your numbers, the names of your contacts, or your internal constraints. Finally, review and adjust every output before sending it: this is the rule the ChatGPT documentation also emphasizes for any important work.

Your turn

Take inventory of your week. Spend five minutes on this exercise before continuing the course:

1. List the ten tasks you do every week
2. Filter for those with a clear outcome: document, email, spreadsheet
3. Rank them by frequency multiplied by time spent
4. Choose the top item on the list: this is your ideal candidate

Then open ChatGPT and test this formula on the selected task:

I need to [task] every [frequency]. Here is the context: [context].
Here is this week's information: [data].
Produce [expected result] in the following format: [format].

Official sources: New OpenAI Academy courses for the next era of work (OpenAI, June 12, 2026) · Prompting (ChatGPT documentation)

Key takeaways

• Start with recurring tasks that have a clear outcome, not with an ambitious project
• Three criteria: recurrence, defined outcome, sufficient expertise to judge the output
• Best candidates: meetings, status updates, sorting requests, minutes, follow-up emails
• Start with one task, master it, then expand; measure time saved rather than assuming it