Korva is a scroll of the internet’s best ideas.
Divorce as a service went viral
From viral post to lawyers
Aparti began as a do-it-yourself product for consumers, something like TurboTax for divorce. The viral post brought a flood of people asking what a divorce would cost them, but the strongest response came from lawyers. Intake is admin work that goes back and forth for months, and it eats the time they could spend on actual lawyering.
Cutting intake time
Intake used to take around 10 hours, spread over months. With a conversational AI that asks a few questions and fills in court-ready forms, it takes about one hour, and the client can answer at their own pace.
The asset division engine
That opened a bigger target: asset division, where lawyers download bank statements, pass them to accountants, and track the outcome by hand. The team is building an asset division engine with a settlement simulator that lays out different outcomes, trained on 140,000 California divorce cases. Nine family lawyers have signed on as design partners, and their data is being used to tailor the product.
Why did lawyers respond more than consumers?
A lawyer who saves hours on intake can bill those hours to other clients or take on more cases, so the benefit is concrete and recurring. A consumer uses a divorce product once, which makes it hard to build a business on repeat use. Businesses that sell to professionals often have higher contract values and lower churn than consumer products. Many startups that begin by serving individuals end up selling to the professionals those individuals would have hired.
How can AI fill in court forms?
Court forms ask for the same facts in fixed fields, such as names, dates, and financial details. A conversational model can gather those facts through ordinary questions and map each answer to the right field. Language models are well suited to this because they turn loosely worded text into structured data. The output is still reviewed by a lawyer before filing, since the lawyer remains responsible for the document.
What does a settlement simulator do?
It estimates how a dispute might end by comparing it to the outcomes of many earlier cases. Such tools are trained on past rulings and settlements, then show a range of results rather than a single prediction. Litigation analytics companies have sold similar outcome forecasts to law firms and insurers for years. The value is in letting both sides see the likely range before spending months arguing.
More from Aparti
