You can turn AI models like ChatGPT and Claude into personal productivity coaches with your Lifeline data — create charts, receive thoughtful feedback, and get insights on your habits and work patterns.

I’ve spent the past weeks putting together a collection of queries for your data. You'll find them organized in this Notion database. It's publicly viewable and, with a Notion account, you can leave comments and even add your own queries!

Here's how it works:

1. Export your Lifeline data to CSV from the Review stats and activities… menu.
2. Attach the CSV to a chat in ChatGPT or Claude.
3. Use a question from the database to analyze your data.

Make sure to check out the video above for more details. And the more you track and label your sessions with Lifeline, the more interesting insights you’ll get from this approach.

Three examples to get you inspired and started

Here are three examples of instructions you can give and the output they generate.

“What is the total number of Pomodoros completed on a week-by-week basis?”

This request for ChatGPT gives you a simple bar chart that shows how many pomodoros you’ve completed for every week in the period covered by the data set.

‍“Create a heatmap displaying the number of Pomodoros completed by the day of the week and the hour of the day on a cumulative basis for the entire period.”

Put this into ChatGPT, and you get a handy visual representation that shows on which days and hours you earned the most pomodoros.

‍“What is the impact of the first work session's start time on the day's productivity?”

Here’s an excerpt of the response Claude gives when asked this question:

"The most productive days often start between 7:30 AM and 8:30 AM. Many of the days with 12+ completed sessions began in this time range.

Very early starts (before 7 AM) don't necessarily lead to more productivity. These days often have long breaks after the initial session, suggesting the early start might not be sustainable.

Consistency seems to matter. Weeks with consistent early starts (like the first week of July) tend to be more productive overall than weeks with variable start times."

Tips to get better results from your analysis

A few things to keep in mind when you analyze your Lifeline data with AI:

• ChatGPT is great for creating diagrams and hard number crunching.
• Claude’s 3.5 Sonnet model performs better than ChatGPT as a coach and analyst.
• Claude is less precise and sometimes makes mistakes, so you need to stay on your toes.
• Attach less data if you want to zoom in on a specific period. You can also export your data without breaks and untimed activity, reducing the size of each file.
• Providing extra context at the start of a conversation is usually helpful. For example, that you were on vacation a certain week or how you use emoji labels.

Help me improve and expand the question database!

The list of questions and instructions in the Notion database is far from complete, and the current ones can use testing and refining. Don’t hesitate to jump in and leave comments on how they work for you. You can also enter your own — I’d love to see what use cases you find for your Lifeline data x AI.

Try the latest version of Lifeline now from the macOS App Store.