Last updated:
Quick answer
An agent saying it will watch for something is not the same as a scheduled job doing it. Turn every promise into a dated task, have a daily check list promises due today and their status, and treat anything not confirmed as not done.
Free PDF
Get this setup as a PDF
Confirm your email and we'll open a print-ready copy of this setup you can save as a PDF. You also get the free starter pack and one setup a week. Unsubscribe anytime.
Already a subscriber? Unlock it on this device
At a glance
All outcome figures are estimates from cited sources, not independent verification by us.
- What it is allowed to do
- Read-only
Only reads, researches, or summarizes. It never drafts outbound content or changes anything.
- Time saved (estimate)
- About 5 minutes a day.
- Difficulty
- Intermediate
- Tools needed
- ChatGPT, Claude
- Typical cost
- Scheduled tasks and connected apps depend on your plan. Prices: Not stated in sources.
Reviewed Nov 2, 2026
Basis (estimate): Estimate only: based on the steps above, not timed.
What does this setup do?
Turn every follow-up your AI agent promises into a dated task, run a daily check of what was due, and get a short list of anything that slipped.
Who is it for?
Owners whose agent said it would watch for a reply or chase something later, and then missed it.
How do you set it up?
Tools
- ChatGPT or Claude
- A simple promise log: a note, sheet or task list
- Your inbox, to confirm replies
Steps
- Each time the agent promises a later action, have it write one line to a promise log: what, for whom, due date.
- Do not rely on the agent remembering. OpenAI documents scheduled tasks in ChatGPT as one-time or recurring with notifications; use a real schedule or a calendar reminder for anything that must happen.
- Each morning, paste the promise log and what actually happened, such as replies received and messages sent, into the prompt below.
- Handle anything marked SLIPPED yourself, or tell the agent to draft it now for your approval.
- If the same promise slips twice, remove it from the agent's job and put it on your own list.
- Review the log weekly with the weekly agent review setup.
Example output
Illustrative. Your results will differ.
- SLIPPED: watch for Dana's reply on invoice #1042 (due Nov 1); reply arrived Oct 31, no answer sent
- Next step: draft a reply to Dana confirming the payment date
- COMING UP: follow up with Lee on the roof quote (Nov 4)
Where to paste this prompt
ChatGPT → Project instructions, or ChatGPT Tasks for scheduled reminders (once a day on Free)
Copy-paste prompt
Adapt bracketed fields to your situation.
Here is my agent's promise log and what actually happened. Use only what I paste.
Promise log (what, for whom, due date): [PASTE]
What happened (replies received, messages sent, tasks done): [PASTE]
Today is: [DATE]
List:
1. DONE: promises with clear evidence they happened
2. SLIPPED: due today or earlier with no evidence
3. COMING UP: due in the next 3 days
For each SLIPPED item, draft the one-line next step. Treat anything without evidence as not done.Newsletter
Get the free starter pack
Free starter pack with five beginner copy-paste prompts after you confirm, plus one researched setup in your inbox each week. Unsubscribe anytime.
What usually goes wrong?
Where agents tend to stall on this job. More patterns in Where AI gets stuck.
- Trusting a chat promise with no schedule behind it.
- Logging promises without dates.
- Letting the agent mark its own work done without evidence.
Which chat tools work with this prompt?
Paste the prompt into any of these assistants (free or paid tiers).
- ChatGPT
- Claude
Common questions
- Why did my AI agent say it would follow up and not do it?
- A chat promise is not a schedule. Unless a scheduled task or reminder exists, nothing guarantees the agent comes back to it.
- How do I make sure follow-ups happen?
- Log every promise with a due date and run a daily check that compares promises to what actually happened.
- Can the agent check itself?
- Only with evidence you can see, like sent messages and replies. Treat anything without evidence as not done.
Where should you go next on this site?
What tips make it smoother?
- Evidence beats memory: a sent message or a logged reply counts, a statement that it was handled does not.
- Put due dates on every promise.
- Keep the log where you can see it, not only inside the agent.
Where do the claims come from?
Summaries use public sources linked below, not independent verification. How setups are picked.
- Read OpenAI’s write-up (help.openai.com)
OpenAI: Scheduled tasks in ChatGPT: one-time or recurring tasks with push or email notifications; project files not available to tasks (read 2026-10-31 through a page fetch tool, not a direct browser open)
- Read OpenAI’s write-up (help.openai.com)
OpenAI: ChatGPT agent: when signed into sites or apps it can access sensitive data and perform actions (read 2026-11-02 from a search result excerpt of the page)
Related setups
- Review what your AI agent did each week
Spend 15 minutes a week reviewing what your AI agent did: sort its actions into fine, fixed and should-not-have-happened, then tighten one rule for each problem.
- Make your AI agent stop for approval
Make your AI agent stop and show you every email, booking or post as a draft with a one-line summary, and wait for your OK before it goes out.
- Set guardrails before an AI agent gets access
Write a one-page rulebook for an AI agent before it touches your accounts: what it can read, what it can only draft, what always needs your approval, and what it must never do.