| |

Using AI to Perfect Your Inventory & Appraisement

Using AI to Perfect Your Inventory

Building a JSON Document for Consistent Presentation

Here is the prompt I use to prepare my AI of choice to help me create an I&A database from which we build settlement proposals, trial exhibits, and formal Inventory & Appraisement documents.

We are going to create a list of json objects for each asset with different properties for each class of asset as follows:

# JSON Specification

# ASSETS

## RetirementAccounts
* institution
* account_number
* name_on_account
* balance
* balance_date
* account_type

## GoldSilver
* form (e.g. US Dollars, Canadian Silver, etc.)
* possession_of (Michelle or Garrett)
* ounces
* price_per_ounce
* price_date (leave this blank for now, I'll fill it in later...unless you can get a current commodity price)

## Financial Accounts
* institution
* account_number
* name_on_account
* account_type
* balance_date
* balance
* function

## Houses
* address
* county
* legal_description
* mortgage_company
* fair_market_valuie
* mortgage_balance

## Life Insurance
* issuer
* policy_number
* insured_name
* owner_name
* type
* face_amount
* current_cash_surrender
* beneficiary_name
* owner_name
* loan_flag (Y or N)

## Vehicles
* possession_of (e.g. Joe, Sally, Preston, Chloe)
* year
* make (e.g. Ford, Subaru, Cadillac)
* model (e.g. Escalade)
* fair_market_value
* valuation_date
* creditor
* loan_balance
* loan_balance_date (if you can find it, otherwise blank)
* vin

## Businesses
* name
* address
* corporate_form (e.g. Domestic LLC)
* formation_date
* ownership_percent
* fair_market_value
* source_of_value
* valuation_date (if you can tell)

## Personal Items
* description
* location ("McKinney House")
* awarded_to ("Joe")


# DEBTS

## Credit Accounts
* creditor
* type (e.g. credit card, student loan, auto loan, home mortgage)
* account_number
* name_on_account
* balance (i.e. amount currently owed)
* balance_date

## Taxes Owed
* taxing_authority (e.g. "IRS", "Collin County Tax Assessor")
* tax_year
* amount_owed

If you can't find a value, don't skip the field, just leave it with a blank ("") or zero value

This sets up the form of the detailed entries in the database. Then I tell the AI how to organize the items.

The json items will be arranged in a database like this:

json
{
"community_estate": {
"assets": [ detailed list of asset objects ],
"debts": [ detailed list of debt objects ],
},
"separate_estates": {
"husband": [ detailed list of asset objects ],
"wife": [ detailed list of asset objects ],
},
"childrens_assets": [ detailed list of children's assets ]
}
```

Next, I start dragging in the most recent statement for each account:

I have attached the most recent statement for each banking account. Review the first page of each PDF and create COMMUNITY_ESTATE - ASSETS - FINANCIAL ACCOUNT objects based on each statement.

Then, drag in the credit card statements:

I have attached the most recent statement for each credit card account. Review the first page of each PDF and create COMMUNITY_ESTATE - DEBTS - CREDIT ACCOUNT objects based on each statement.

And now for the payoffs. Generate a settlement spreadsheet:

Create a settlement proposal spreadsheet with these sections:

# Husband's Proposed Property Division
## Community Estate
### Assets
### Debts
## Husband's Separate Estate
## Wife's Separate Estate
## Children's Property

For each row, include the following columns:

1. Description
2. Value
3. Possession Of
4. To Husband
5. To Wife
6. Notes

For now, leave the "To Husband" and "To Wife" columns blank.

After the Ai generates the spreadsheet, I print the spreadsheet and then check each line item against each source document. This database is the foundation of all that comes next, so you want to make certain that it is absolutely perfect.

If I find changes that need to be made, I tell the AI to make the changes, e.g.:

I have some edits for the database:

1. Change "Wells Fargo x1234" to "Wells Fargo x2134"
2. Change "VISA x9878" so taht it is in the name of "John Doe and Sally Doe"
3. Add a bank account:
* institution: Bank of America
* account_number: x1432
* name_on_account: John Doe
* account_type: Savings
* balance_date: 06/26/2026
* balance: #14,325.98
* function: General savings

After you do that, update the spreadsheet.

Do another print/verify step. Iterate until you have no more changes. Next, try doing a real property division:

Let's now assign community assets and debts between parties. Start by assigning everything that is husband's name or possession to husband and the remainder to wife. Once you've done that, regenerate the spreadsheet.

Look at the resulting property division. This will be the easiest property division possible: Everyone keeps their stuff. That probably won’t work. To come up with a workable solution, how much you have to move from one column to the other to get to a 50:50 division:

What is the dollar amount in assets that we need to move from wife's column to husband's column to achieve a 50:50 division of the community estate. Once you find that total, try to identify an account or set of accounts that can either be moved or divided between parties and give me a report of those.

Keep working with it until you have a workable spreadsheet.

Next, create the formal Inventory & Appraisement. To do that, find an inventory that has a similar type of properties as this estate. FInding it in PDF form is fine.

I have attached an "Inventory & Appraisement" from another case. Use the format of that Inventory to create an inventory for this case. I have also attached a pleading that shows the complete style of the case. We represent "John Doe."

When you see errors or omissions in the data, fix them in the JSON database as shown about. By making corrections and additions to the JSON database, you will always be able to regenerate spreadsheets and inventories with consistent, correct data, driven from a single source of truth.

Leave a Reply

Your email address will not be published. Required fields are marked *