A Claude prompt that looks at your last seven days and gives you three specific things AI would have made better.
A prompt is just text you paste into a chat to tell it what you want.
A connector lets Claude read from apps you already use, like email, calendar, or Slack. Instead of you copy-pasting context into a chat, Claude can pull it directly. You set it up once and Claude has access in every conversation after. Set one up here.
Copy the prompt below into a fresh chat. Don’t edit it the first time. Run it as written, then adjust.
Look at my activity from the past 7 days across the apps I have connected (Gmail, Calendar, Slack, Notion, Drive, anything you can access). First, get a feel for how I already use AI based on what you can see. Assume I’m not technical and don’t want to build anything. I want things I can do myself, in a chat window, starting tomorrow. Find 3 specific moments from this week where AI could have changed the outcome of my work, not just done it faster. Look for: - Things that crossed multiple apps (a meeting that became emails and notes and Slack threads) - Decisions I made where more context would have helped - Work where the result could have been meaningfully better - Things I might not realize AI can help with For each moment: - Describe what I actually did, with a specific example from my data - Explain what AI could have added (better context, a catch I missed, a draft worth keeping, an angle I didn’t see) - Give me one thing I could do next time, by hand, in a chat. A prompt I could save and reuse, or a question to ask before I start. Something that takes a minute, not a project. Do not tell me to build an automation, an agent, or anything technical. Skip surface-level stuff like “summarize this email.” I want patterns I’m not noticing in my own work. If my week was genuinely light, say so plainly and give me fewer than three moments rather than stretching to fill the list. One real thing beats three forced ones. At the end, name the pattern. What does the shape of my week tell you about where AI most naturally fits into how I work? Then pick the single most useful of the three and offer to do it with me right now, using my real data, so I can see what it looks like.
The first time you connect a few apps to Claude, the most useful question is not “what can it do.” It’s “what could it have done last week.”
This prompt looks backward at real work you already finished. The result is concrete, personal, and a little uncomfortable in a good way. Two connectors is enough to start, more makes it sharper.
You had a 45-minute meeting on Wednesday. Real questions came up. Your follow-up email the next day said “thanks for the great conversation” and referenced the topic in general terms, but didn’t address the specific concern the other person raised at minute 12. AI could have pulled that point from the meeting notes and worked a real answer into your follow-up.