Build a personal reading profile from exported history
At a glance
A writer used ChatGPT to combine exported activity and an interview into a profile for more personal book suggestions. This workflow is listed under Reading, Personalization, Learning.
- Agents / tools
- ChatGPT, Google Takeout, Reddit
SourceBuilding a tool for these workflows? See weekly rates.
What they did
1) Request a Reddit data export and download selected Google Takeout records. The writer used years of comments and discussions along with search and YouTube activity. 2) Create a ChatGPT Project, upload the material, and ask it to identify recurring themes. 3) Ask ChatGPT to interview you about the data so it can add the reasons behind the interests; after roughly 20 minutes, it produced a document covering biographical details, preferences, recurring themes and communication style. 4) Tell the Project to respond as the person described by the documents and interview, with the profile open to later updates. Test it with a book recommendation and review the result against your own taste.
How it went
The author received a book suggestion he had not mentioned in the interview or on Reddit and added it to his reading plans.
Worth knowing
The writer noticed odd extrapolations and an exaggerated interest in puns. The source describes an experiment in personalization, not a faithful copy of a person.