chevron-down Created with Sketch Beta.
February 03, 2020 Feature

Imagining an AI-Supported Self-Help Portal for Divorce

By Alan Carlson

Chris and Sam have been married for 10 years. Lately their marriage has just not been working for them. After talking with friends and trying counseling, they decide that they want to get a divorce. Being millennials, they turn to their phones to find out what is involved in getting a divorce and to find a do-it-yourself divorce website. “It shouldn’t be too complicated, right?” On their phone’s web browser they enter “how to get a divorce.” What sort of results will their search reveal?

What Is Available Now

Today the search results may include several types of assistance. Some websites provide basic information about divorce, some quite detailed.1 Others offer fillable court forms. Some offer dispute resolution services. Some offer assistance for particular issues, like child custody and visitation disputes. There will also be lawyer referral services or local lawyers willing to represent them for a fee. However, other than the lawyers, currently none of the options noted are able to guide them all the way through the legal process (as opposed to just descriptions of legal issues), negotiating issues they disagree on, and obtaining a divorce judgment signed by a judge. While there are some tools and websites to assist,2 and organizations have been working on this,3 there is not yet a “one-stop shop” website for a typical, not-too-complicated divorce appropriate for a majority of couples seeking divorce.

What Is Possible

Recent advances in technology could provide self-represented litigants who are willing to work together a realistic, practical, online option to “do-it-yourself.” With recent developments in data analytics and tools developed using artificial intelligence (AI), it could be possible to provide more substantive assistance from the beginning (contemplating a divorce) to the end (getting a divorce judgment).

A portal that includes a variety of tools built from an analysis of past divorce cases and that incorporates applicable law and legal procedure could be assembled that could


  • Help people understand what the legal, economic, and practical issues are in a divorce;
  • Help people understand the range of possible outcomes, what is reasonably and realistically likely, and what is not possible;
  • Identify documents and information the parties need to gather to proceed;
  • Help navigate the legal process, in terms of what the next steps are, what they involve, and the timing of each step;
  • Provide document assembly and document review for court filings, proceedings, and judgment;
  • Offer options in addition to “going to court” for resolution of any disputes that come up;
  • Help in scheduling, preparing for, being reminded of, and appearing at court hearings; and
  • Help in finalizing and implementing a judgment.

Such an online toolkit could not help in every case. If the couple is locked in a psychological war or one or both are seeking unreasonable outcomes, such a portal likely would not help. In addition, couples with complicated, unique, or numerous assets; businesses; occupations; children (for example, with special needs); or other complicating factors may still need to engage a lawyer with knowledge and skills related to their unique or less-common issues.

How It Might Work

Back to our couple looking for a divorce. In the future scenario outlined below, one of the websites found by the browser could be a portal that guides the couple through the entire divorce process. The portal would appear as one site, but it would consist of a network of online tools, applications, guides, explanations, and referrals linked together in a manner seamless to the user. The portal would integrate services provided by the court, other government entities, nonprofit and social service entities, and for-profit entities.

The couple would begin by looking at the portal app that provides basic information—explaining what a divorce is and providing an overview of the law about divorce and the legal steps involved—a sort of Wikipedia for divorce. The app could have a chat feature, allowing people to ask questions and seek clarification. In the background, the app would be monitored and, with the assistance of experts, new questions would be answered and information updated to provide a better experience for future users. The objective would be to provide a basic understanding of the divorce process and help each person decide whether to continue. They could refer back to this app at any stage of the process for basic information about what is happening.

Another aspect of the first stage in the portal would be a “what if” tool that a person could use to see what the possible outcomes of divorce are in his or her situation. Currently, only anecdotal information about what happens in divorce is available, usually from family and friends (who often carry biases and misimpressions based on their experiences). With the portal, individuals could enter basic information about themselves, their partner, and their marriage to see what might happen. Based on an analysis of thousands of past divorce cases, the app would then provide a range of possible outcomes for them given their circumstances. AI tools are usually built to provide a single prediction. This tool, instead, would provide a range of possible outcomes, not just the most likely one. Providing a realistic picture of what might happen could avoid surprises and minimize some of the frustrations and impediments from unmet expectations that occur in a typical divorce.

Filing for Divorce

Having digested this information, our couple decides they do want to file for a divorce. Using a question-and-answer format, the portal would next ask each of them to provide more detailed information about themselves and the marriage, from which the portal can determine what legal issues must be resolved. At each step of the legal process, the portal would ask for information needed at that stage. The information would be stored in a secure locker to protect each person’s privacy. The initial information requested would determine if they are eligible for a divorce and what court is the proper venue. Additional information collected at this stage would also include information about any children, sources of income, property and other assets owned by the couple, medical insurance, retirement accounts, and so forth, in order to identify what legal issues will need to be resolved.

Our couple, for example, would indicate where they live and for how long, the date of marriage, and that there is one child, both are employed, one employer provides the family’s medical insurance, there is one defined contribution plan and one 401(k) plan, and they rent their home, own two cars, and have bank accounts and credit cards. Based on this information, the portal would know this couple will have to address legal issues regarding child custody and visitation, child support, division of assets and debts, health insurance coverage, allocation of the retirement plans, and, possibly, spousal support.

Having gathered the needed information, another app on the portal would fill out the necessary court forms for opening a divorce case and send it to the parties for their review. If they decide to proceed, the app would e-file the forms with the court and electronically serve the parties. The parties would receive confirmation and would be able to access the filed documents through links in their information locker. Whether there is a need for temporary orders pending final resolution would also be assessed by the portal, and the parties would be directed to processes similar to those described below.

Working Through the Issues

The next set of portal tools and apps each addresses a different legal issue that must be resolved in their case. These issues could be pursued in parallel; they need not be done seriatim or in any particular order. Focusing on the child, there would be a path for resolving child custody and visitation issues. The tool would ask for additional information not already collected that would be relevant to resolving the child custody issue. For example, does the child have special medical or educational needs? What sort of childcare is currently being used? What is the child’s school schedule, for example? At this point, an online interactive process, using a template, could be used to determine where there is agreement between the parties regarding custody and visitation. When there is a lack of consensus, or ideas about how to arrange custody and visitation, a tool that has analyzed thousands of past cases, annotated by child custody experts, could propose options to fill in the gaps where the couple does not agree or does not know what to do. The parties also could choose an app that provides a referral to a third-party service to advise on a particular issue.

If the parties could not agree about a legal issue, the portal would offer options for resolving the dispute, including mediation, arbitration, online dispute resolution (ODR), or some other third-party-facilitated dispute resolution service, as well as “going to court.” There is no intent to replace the court, only to make it clear there are other options and to provide information about the options. “Going to court” is always available.

For services to be provided by a third party, there would have to be some vetting of the entities providing services, service level agreements, and monitoring of performance, including use of rating agents. If the parties pick a third party, the referral process would happen electronically. The outcome of the referral would be reported back to the portal electronically by the third party. If agreement was reached on issues, the terms of that agreement would be incorporated by the portal into subsequent documents, including the proposed final judgment.

If the parties are unable to reach an agreement, either by themselves or with the assistance of a third party, or if they choose to go directly to court, they could have the court decide outstanding legal issues. There would be another path in the portal for court hearings. When a court hearing is requested, the app would take care of scheduling the hearing by connecting with the court’s case management system and selecting a hearing date acceptable to the parties and the court. It also would request any needed additional information, assist in preparing and filing documents for the hearing, and send required notices for the hearing.

Before the hearing, another app could provide information to help the parties prepare for the hearing and also let each party know what documents each must bring. The court hearing app also could provide hearing reminders to the parties and provide way-finding directions, both to get to the courthouse and to get to the courtroom once inside the courthouse.

After the court hearing, and the judge has made a decision, the court prepares an order reflecting the decision. Once approved, the clerk submits the order to the portal, which would serve the parties. The portal would retain the information in the decision for preparation of the proposed final judgment. If the hearing was recorded, any party could request, through the portal, a copy of the oral or video record of the proceeding. If the proceeding was recorded electronically, AI tools could convert the oral record to a rough transcript when requested.

Getting a Judgment

When the couple have resolved all of their disputes and agreed on the terms, the portal would walk them through assembling a proposed judgment. When this proposed judgment is submitted, it would first be reviewed by an AI-powered reviewing tool. This tool would look for anomalies, incomplete terms, inconsistencies, distributions of assets or support that appear imbalanced, or other indicators of unfairness, etc. While it is currently impossible to build an AI tool that adequately measures the “fairness” of a proposed judgment, an AI tool could compare the proposed judgment to previous judgments to look for “outlier” aspects, unusual patterns, or other inconsistencies. Once it has conducted its review, the review tool would send the proposed judgment along with the tool’s analysis to the judge.

The judge reviews the proposed judgment, considers the tool’s analysis, and either signs the judgment or returns it to the parties with an explanation of the reasons for not signing. The parties can revise the proposed judgment if they can agree; conduct further negotiations, possibly with the assistance of a third party; or request a court hearing. Once the judge approves the agreement, or makes a decision after a hearing, another portal app would prepare the final judgment, get it signed by the judge, and send it to the parties. Another app would tell the parties what must be done to implement the judgment; for example, how to change title to property, tax consequences, and how to notify health insurance or retirement plans.

Post-judgment proceedings could be handled in a similar manner by another path in the portal. It would include apps for filling out and filing needed forms, offering parties alternative dispute resolution options, preparing and reviewing proposed orders, and so forth. The apps would use information that is already available from the locker and would only ask for any new information needed for this stage.

Other Aspects of the Portal

Paraphrasing an old adage: “When you have seen one case, you have seen one case.” There are many other legal issues not described above that can have their own paths, apps, and tools in the portal. For example, there could be a path for requesting to proceed in forma pauperis. Joining of necessary parties, such as retirement plans, could also be another path.

A more difficult path that must be included would be one addressing domestic violence, spousal abuse, and child abuse. From the beginning, the portal needs to be aware of, and ask questions to assess the presence of, domestic violence or other power imbalances that the court needs to address. A basic goal of the portal must be to identify all possible legal issues, including the presence of violence or abuse, and not to rely on the parties to do so.

All information requested by the portal would be kept in a secure information locker. Each party would have his or her own locker and could access or change only the information that individual provided. Links to the court record would also be available through each party’s locker. During each stage of the proceedings, each party could authorize portal apps or tools to access his or her information; for example, to fill out forms, make referrals, and so forth.

Everything in the portal would be done electronically. Legal documents would be prepared electronically, signed electronically, served electronically, and stored electronically in the court record. Similarly, referrals would occur electronically, with relevant information transferred electronically and securely to third parties providing services chosen by the parties. Outcomes of third-party services—for example, a completed or partial agreement on issues—would be returned to the court electronically as well.

The comprehensive service portal proposed above does not currently exist. The portal will require the development of a framework or platform that seamlessly provides the paths, referrals, and apps needed for all of the options. The technology to build the pieces is available. Some of the pieces described already exist. Basic information about the divorce process, divorce law, and court forms already exists but needs to be reorganized so as to be accessible to the portal apps.

The portal would include a variety of specialized tools. Some would require extensive “data analytics”—an analysis and synthesis of data from thousands of divorce cases already concluded. The analysis would identify the range of acceptable alternatives and characteristics of the alternatives for people to consider in deciding whether to get divorced and in resolving their own divorce if they decide to file. Other tools, for example, the review of proposed orders, would require models developed in consultation with subject-matter experts to analyze proposed orders and judgments in light of applicable laws and acceptable outcomes.

Getting There

As always, technology provides tools, not the complete solution. The tools contemplated here work best where there are rules, “right answers,” formalized processes, and patterns discernable in a dataset of past divorces. Current technology is not as adept where the issues are “conceptual, abstract, value-laden, open-ended, policy- or judgment-oriented; require common sense or intuition; involve persuasion or arbitrary conversation; or involve engagement with the meaning of real-world humanistic concepts, such as societal norms, social constructs, or social institutions.”4 Consequently, there will always be a need for humans in the loop—the parties themselves, the people providing third-party services, court staff, and judges. The tools are intended to support the people in resolving what is a very human activity—the reorganization of the relationships of a couple and their family.


1. See, e.g., Alaska Ct. Sys., Self-Help Center: Family Law,; Family Law: General Information, Orange Cty. (Cal.) Superior Ct.,

2. For example, see the portal developed by the Orange Cty. (Cal.) Superior Court self-help center at See also services described in Danielle Braff, Do DIY Divorce Apps Deliver? Services Promise an Easier Process, ABA J. (July 1, 2019),

3. For example, the Institute for the Advancement of the American Legal System (IAALS) Court Compass Project and the Legal Services Corporation LawHelp Interactive website.

4. Harry Surden, Artificial Intelligence and Law: An Overview, 35 Ga. State U. L. Rev. 1305 (2019) (Univ. of Colo. Law Legal Studies Research Paper No. 19–22, 2019).

The material in all ABA publications is copyrighted and may be reprinted by permission only. Request reprint permission here.

Alan Carlson

Alan Carlson retired as the court administrator, clerk of court, and jury commissioner of the Orange County, California, Superior Court. He is now a consulting lawyer/engineer.