Using AI to help separating couples split assets fairly

Summary

Funded by the Australian government and initiated by Legal Aid, Amica uses AI to help amicably separating couples reach fair property and parenting agreements without a lawyer. Launched in 2020 and backed by the Attorney General, it has helped over 9,000 couples reach agreements.

We worked with family lawyers to train the model on real lawyer recommendations, then designed a plain-language flow that guides couples through joint statements, turning them into a court-ready agreement, with education and sentiment analysis built in.

Role

One of the first and lead product designers, working closely with data engineers and subject matter experts.

Using AI to help separating couples split assets fairly

Summary

Funded by the Australian government and initiated by Legal Aid, Amica uses AI to help amicably separating couples reach fair property and parenting agreements without a lawyer. Launched in 2020 and backed by the Attorney General, it has helped over 9,000 couples reach agreements.

We worked with family lawyers to train the model on real lawyer recommendations, then designed a plain-language flow that guides couples through joint statements, turning them into a court-ready agreement, with education and sentiment analysis built in.

Role

One of the first and lead product designers, working closely with data engineers and subject matter experts.

Using AI to help separating couples split assets fairly

Summary

Funded by the Australian government and initiated by Legal Aid, Amica uses AI to help amicably separating couples reach fair property and parenting agreements without a lawyer. Launched in 2020 and backed by the Attorney General, it has helped over 9,000 couples reach agreements.

We worked with family lawyers to train the model on real lawyer recommendations, then designed a plain-language flow that guides couples through joint statements, turning them into a court-ready agreement, with education and sentiment analysis built in.

Role

One of the first and lead product designers, working closely with data engineers and subject matter experts.

A lot of the feedback received was around how helpful it was to view the process step by step in plain conversational language.

Working closely with family lawyers, the user flow was based on the order they typically follow. Questions are asked one at a time to reduce cognitive load and provide additional context and clarification if needed. The language is conversational in nature and applied contextually throughout to provide support and explanations along the way.

Starting as a proof of concept, the federal government then provided funding to investigate how online dispute resolution could offer an affordable solution for separating couples.

Disputes that arise from the breakdown of a relationship can be particularly expensive, especially for those who don’t qualify for legal aid.

Impartial, always

Every interaction is designed with fairness at its core, ensuring all parties feel equally valued and respected, free from bias or intimidation.

Impartial, always

Every interaction is designed with fairness at its core, ensuring all parties feel equally valued and respected, free from bias or intimidation.

Clear at every step

Information appears when needed, helping users navigate confidently without feeling overwhelmed by complexity or legal jargon.

Clear at every step

Information appears when needed, helping users navigate confidently without feeling overwhelmed by complexity or legal jargon.

Expertise with empathy

We combine legal knowledge with human understanding, making complex matters feel both trustworthy and approachable.​​​​​​​​​​​​​​​​

Expertise with empathy

We combine legal knowledge with human understanding, making complex matters feel both trustworthy and approachable.​​​​​​​​​​​​​​​​

Creating a pictogram set to create a clear and approachable separation between key sections.

Feedback highlighted the importance of creating a more collaborative approach to reviewing statements to help separating couples move through the process.

Fairness was the foundation throughout the design of the negotiation process, as users elect to either agree to the AI powered suggestion or use it as a guide.

AI provides a suggestion of what a lawyer would consider a fair division of their assets based on the information they’ve entered. Instead of using published case data to power the machine learning algorithm, (which the team found contained biases), the team created there own data model that lawyers contributed to.

Hints and alerts provide guidance if you attempt to make an unfair offer (outside the bounds of the prediction).

Featured on talk shows and ABC news

9,000+

Separating couples have used Amica. People around all of Australia can use Amica instead of paying to see a lawyer.

9,000+

Separating couples have used Amica. People around all of Australia can use Amica instead of paying to see a lawyer.

$80M+

Saved the courts in out of court agreements.

$80M+

Saved the courts in out of court agreements.

Launched February 2020

Now part of a coordinated effort across all legal aid jurisdictions to refer clients.

Launched February 2020

Now part of a coordinated effort across all legal aid jurisdictions to refer clients.