Filing out all the paperwork for a new apartment or home mortgage can be tough enough, but there’s one gotcha that can be particularly annoying: detailing the past addresses where you’ve lived over the past five, 10, or even 20 years, including dates for when you moved in and out.
That tripped me up last year as I was completing a co-op application, leaving me to dig through my files for leases dating back a decade or longer. When did I move into that Brooklyn apartment on Henry Street, and what was the apartment number again?
I got my address history ducks in a row eventually, but there’s an easier way to accomplish it—with the help of an AI agent.
For this task, we’ll be using Gemini’s Spark agent and its Gmail connector. (Yes, you’ll need to allow Spark to comb through your Gmail inbox.) If you’re not comfortable with that, no problem—feel free to skip this tip. If you’re new to Gemini’s Gmail connector, see this article first for details about the best configuration settings and what you need to know about privacy.
Once you have Spark’s Gmail connector configured, start a new thread and give it this prompt (which was crafted by Claude Opus 5.5):
I need my address history for a form that asks where I’ve lived for the past [X] years. Search all of my Gmail, including archived mail, from [START YEAR] to today.
1. Look for anything that shows where I lived and when: leases and rental applications, move-in/move-out and security deposit emails, utility and internet start or stop services, USPS change-of-address confirmations, DMV or voter registration confirmations, mortgage or closing emails, insurance policies, and changes in the “ship to” address on my own orders.
2. Create a new spreadsheet called “My address history” with one row per address, newest first: full street address, city, state, ZIP, earliest month I lived there, last month I lived there, how confident you are (confirmed by a lease/closing/USPS email, or inferred from orders/bills), and links to the 2–3 emails that prove each date.
3. Flag any gap or overlap of more than one month between addresses, and list what kind of email would settle it.
Only use addresses and dates you can link to an email. If you can’t pin a date down, write “unknown” rather than guess. Don’t include anyone else’s address (gifts, people I shipped to). Don’t fill in or submit any form, sign in anywhere, or contact anyone.
I tested this prompt with my own Gmail account, which has messages stretching back more than 20 years. Within a few minutes, Spark spat out a spreadsheet that looked reasonably accurate, piecing together my rental history stretching back to 2006 using everything from lease agreements and move-in notices to security deposit returns and utility confirmations.
The spreadsheet also specified which dates Spark was confident about and which ones it wasn’t. For example, it inferred the last month I lived at my oldest address based on various orders and bills that it found. That’s not iron-clad proof that I moved out in a given month, but it’s a solid guess—and more importantly, I know it’s a guess, not fact.
This agentic AI prompt could work with a number of other agents, too, including a ChatGPT “dot” agent, Claude Cowork, and Meta Muse (and, in fact, this prompt was inspired by this article about Muse). I tested the prompt with ChatGPT, Claude, and Muse agents and got similar results from all of them, but Spark’s spreadsheet was the most thorough.