Automation Recipe: Assemble a Draft Lender Package Before You Even Schedule the Call

What This Builds

A refinance conversation usually starts with a scramble: pull the rent roll, dig up last year's P&L, write a narrative explaining why the numbers look the way they do. This automation starts that scramble for you ahead of time. On a schedule, or whenever you drop fresh exports into a watched folder, it drafts the narrative and assembles a Google Doc packet that's most of the way finished before you've even called the lender.

Prerequisites

• A Professional account: zapier.com
• A Plus account connected through Zapier's AI action
• A Business Standard account for Google Drive and Docs
• A Google Drive folder where you already save your monthly rent roll and profit-and-loss exports from your accounting platform
• A rough sense of your standard lender package structure (property summary, rent roll, income and expense statement, narrative)

The Concept

This automation works like a paralegal who watches your filing cabinet. The moment a new rent roll or P&L export lands in a specific Drive folder, it reads the numbers, drafts the narrative paragraph lenders expect (why occupancy dipped, why an expense spiked, what's changed since last year), and drops the whole thing into a new Google Doc. You review it, fix anything that reads off, and send it. You don't start from a blank page every time a lender asks.

Build It Step by Step

Part 1: Set Up the Trigger

Log into zapier.com and click Create Zap. Search for Google Drive and choose the New File in Folder (Realtime) trigger. Connect your account and select the folder where your accounting platform exports land, or where you manually drop your monthly rent roll and P&L files.

If you'd rather trigger this from a form instead of a folder, use Google Forms → New Form Response with a short form you fill out yourself each time you're preparing for a refinance (property name, reason for refinance, anything unusual to mention).

Test the trigger with a real export or form submission.

Part 2: Draft the Narrative

Add an action step. Search for ChatGPT (OpenAI) and choose Analyze Text. In the Text to Be Analyzed field, insert the file content or form response from your trigger. Set Model to GPT-5.6 Sol.

For the analysis instructions, use a prompt close to this:

You are drafting the narrative section of a lender package for a refinance
on a residential rental property. Using the rent roll and profit-and-loss
data below, write a narrative of 200 to 300 words covering:
- Current occupancy and any vacancy, explained plainly
- Year-over-year change in income and major expense categories
- Any one-time expenses that make this year's numbers look worse than
  the property actually performs (e.g. a roof replacement)
- A closing sentence stating the property's current debt service coverage
  trend in general terms, without restating exact dollar figures already
  shown elsewhere in the package

Keep tenant names out of the narrative entirely. Refer to units by number
only. Write in a factual, professional tone a loan officer would expect,
not a sales pitch.

Data:
[insert file content or form response]

Test and review the draft narrative for tone and accuracy before moving on.

Caution: this workflow moves your rent roll and profit-and-loss numbers through Zapier and ChatGPT on their way to a lender. Keep tenant names out of every file that feeds the trigger. Unit numbers are enough for a lender to follow. Never let a personal financial statement, tax return, or bank statement with account and routing numbers pass through this Zap. Assemble those by hand and attach them separately, and check Zapier's and ChatGPT's current data retention settings before you connect an account that will handle this on a schedule.

Part 3: Assemble the Google Doc

Add another action step. Search for Google Docs and choose Create Document From Text. Set the Document Name to something dynamic, like "Lender Package Narrative: [property name] [date]" using fields from your trigger. In Document Content, insert the narrative ChatGPT drafted, and set the Drive Folder to a "Lender Packages" folder you check regularly.

Part 4: Notify Yourself

Add a final Gmail → Send Email action to yourself with a link to the new Google Doc, so you know a draft is ready to review rather than having to check the folder manually.

Turn the Zap on.

Real Example: The Spring Refinance

Setup: An investor drops the March rent roll and year-end P&L export into a watched Drive folder ahead of a planned cash-out refinance on a fourplex.

Input: The two exported files land in the folder within minutes of each other.

Output: ChatGPT drafts a narrative noting that one unit sat vacant for six weeks during a renovation, that a water heater replacement inflated maintenance costs for the year, and that income has otherwise trended upward. A new Google Doc titled with the property name and date appears in the Lender Packages folder, and an email notification lands with the link.

Time saved: The investor still assembles the full packet by hand (bank statements, tax returns, the actual rent roll spreadsheet), but the narrative paragraph that used to take an hour of staring at a blank page is already drafted and waiting for a quick edit.

What to Do When It Breaks

• The narrative reads generic instead of specific to the property → The prompt probably didn't get enough detail from the file. Check what Zapier actually extracted from the export (some P&L exports are formatted as images or scanned PDFs that don't parse as text). Switch to pasting the relevant numbers into the trigger form instead if the automatic file read keeps coming back thin.
• The Zap silently stops picking up new files → This is the one you won't notice until a lender asks for the packet and nothing's waiting. Check the Zap's history page monthly, or set a recurring reminder tied to your normal month-end bookkeeping close, since that's when you'd expect a new file to trigger the Zap anyway.
• ChatGPT includes a tenant's name → Tighten the prompt with a harder instruction: "If any tenant name appears in the source data, replace it with the word TENANT and the unit number only." Then spot-check the first few runs by hand until you trust it.
• The Google Doc lands in the wrong folder or with a messy name → Double-check the Drive Folder and Document Name fields in the Create Document From Text step. A dynamic field left blank falls back to a generic name that's easy to lose track of.

Variations

• Simpler version: Skip the Drive folder trigger and use a Google Form you fill out by hand right before you start assembling a package. Less automatic, but easier to control what data goes in.
• Extended version: Add a step that also pulls your prior year's narrative from the same Drive folder and feeds it to ChatGPT alongside this year's numbers, so the draft can note year-over-year changes in its own words instead of you supplying them.

What to Do Next

• This week: Build the form-triggered simpler version and test it with a real (or practice) refinance scenario.
• This month: Switch to the folder-watching trigger once you trust the narrative quality, so it runs without you remembering to fill out a form.
• Advanced: Chain in a step that also assembles a cover-letter-style summary email to the lender, using the same narrative as a source, so the whole first contact is drafted in one pass.

Advanced guide for real estate investor professionals. These techniques use more sophisticated AI features that may require paid subscriptions.