Your Family Court Case Needs a System: CRM, AI and Evidence for Litigants in Person

The short answer

A family-court case can generate applications, orders, witness statements, Cafcass material, police disclosure, medical records, emails, screenshots, messages, position statements, chronologies, hearing notes and deadlines. The problem is not always that information is missing. Often, it is that the information exists in too many places and cannot be retrieved, checked or connected quickly.

A litigant in person therefore needs more than a folder of documents. They need a simple personal case-management system which can answer, at any time:

• What is the current order?
• What must happen next, by whom and by what date?
• What issues does the court actually have to decide?
• Which facts are alleged, admitted, disputed or already determined?
• What source document supports each important proposition?
• What has been filed, served, received or merely drafted?
• Which version is authoritative?
• What material may lawfully and safely be shared?

This does not require Salesforce, a legal-practice platform or a complicated collection of apps. For one case, the best solution may be a secure cloud folder, a well-designed spreadsheet, a calendar, a competent PDF tool and carefully controlled use of AI.

AI should sit on top of an organised case record. It should never become the case record.

From business CRM to a personal case system

This article was prompted by Sarah Riesenberg’s Protiviti Technology Insights article, Beyond CRM: Architecture Firm Discovers an Integrated Operating Model for AI-Driven Growth, which I discovered through a LinkedIn post shared by Adrian Anglin, Vice President of Business Development at Protiviti, after we connected on LinkedIn.

Riesenberg’s article is about an architecture firm, Salesforce modernisation and revenue growth—not litigation. But its underlying lesson travels well. The firm’s difficulty was not simply an outdated platform. It was fragmented processes, duplicated information, inconsistent definitions and disconnected systems. Protiviti’s central point is that AI readiness depends upon the operational and data foundations built before AI is introduced.

Family litigation presents a human version of the same systems problem. Critical information may be divided between a phone, an email inbox, downloads, several cloud drives, paper files, messaging platforms and a person’s memory. Dates may be recorded differently. The same document may exist in several versions. A screenshot may be separated from the conversation which gives it meaning. A deadline may be buried in paragraph 17 of an order.

The answer is not to buy more technology. It is to create a consistent operating model for the case and then choose the smallest number of tools capable of supporting it.

What “CRM” should mean for a litigant in person

CRM ordinarily means customer relationship management. A litigant does not have “customers” within a family-court case, so the sales language is not a comfortable or useful fit. What is useful is the discipline behind it: maintaining a structured record of people, communications, events, tasks and documents rather than relying upon memory and an overflowing inbox.

For litigation, it is better to think of the CRM layer as a case hub or case command centre. It is the control panel for the matter. It does not decide the case and it does not contain every page of evidence. It tells you what exists, why it matters, where the authoritative copy is held and what action follows.

A person with a single case will often be better served by a carefully designed spreadsheet than a dedicated commercial CRM. A database-style workspace becomes useful where the case has many linked people, issues, orders, communications and evidence items. Complexity should be earned by a real need; it should not be added because a platform looks impressive.

The central design rule: one source of truth

The most important decision is not which app to buy. It is which location contains the authoritative version of each kind of information.

A sensible model is:

1. Original evidence store: complete and unaltered source material.
2. Case hub: structured metadata and links pointing to those sources.
3. Working area: copies used for highlighting, OCR, extraction, drafting and analysis.
4. Filed and served record: the exact versions actually filed or served, with proof of transmission.
5. AI workspace: selected and, where appropriate, minimised or redacted working material—not an uncontrolled copy of the whole case.

Do not paste the full contents of documents into multiple notes, spreadsheets and AI chats. That creates stale versions and makes corrections difficult. The case hub should ordinarily hold a short description, structured fields and a direct link to the authoritative file.

The practical litigation tech stack

There is no single “best” product combination for every litigant. A safe and sustainable stack is one the user can understand, afford, export and maintain under pressure. Wherever possible, choose one main ecosystem and avoid creating unnecessary links between six different services.

Product names above are examples, not endorsements. Features, prices, AI settings and data practices change. Before using any service for court material, check its current terms, access controls, retention, training choices, sharing permissions, deletion and export arrangements.

JSH Law’s Legal Tech Test series considers legal-technology tools through an access-to-justice lens, including usability, evidential integrity, privacy, cost and practical value for people without lawyers.

The simplest stack is often the strongest

For many litigants, a Microsoft-centred system—Outlook, OneDrive, Word, Excel and Calendar—or a Google-centred system—Gmail, Drive, Docs, Sheets and Calendar—will reduce duplication because the basic tools already work together. A mixed system can work, but every additional integration creates another permission, copy and possible point of failure.

You almost certainly do not need Salesforce. You may not need a dedicated CRM at all. The correct question is not “What is the most powerful platform?” It is “What is the least complicated system that lets me preserve, find, understand and verify my case material?”

What to record in the case hub

A good case hub is not one enormous spreadsheet with no internal logic. Use separate tabs or linked tables for distinct kinds of information. Each record should have a stable ID and a link to the supporting source.

1. Case dashboard

• case name and number;
• court and contact details;
• type of proceedings;
• latest order and date;
• next hearing, venue, format and time estimate;
• next three deadlines;
• current applications;
• issues the court is being asked to determine;
• order or directions sought at the next hearing; and
• links to the latest chronology, position statement and working bundle index.

2. Orders and directions register

Never reduce an order to a vague task such as “send statement soon”. Record the exact wording, paragraph number, responsible person, due date and proof of completion.

3. Chronology

Record the date, neutral event description, source ID and page, issue, evidential status and significance. Keep adult emotion out of the event description. If a date is approximate or disputed, say so.

For a detailed chronology structure, see JSH Law’s Family Court Chronology Templates and Drafting Guide.

4. Evidence register

Give each evidence item a stable reference such as EV-001. Useful fields include:

• date and time created or sent;
• document or media type;
• author, sender, recipient or source;
• short neutral description;
• issue or allegation to which it relates;
• status: alleged, agreed, admitted, disputed, not determined or found by the court;
• original filename and storage link;
• whether the full context and metadata have been preserved;
• confidentiality or sharing restriction;
• whether it has been filed, served, admitted or included in an approved bundle;
• bundle reference and page, if applicable; and
• verification date and any problem requiring attention.

5. Issues and allegations matrix

Separate the issue from the evidence. One row should identify the allegation or issue, the response, the supporting and contradictory sources, whether a finding has been made, and why the matter is relevant to the legal decision. An allegation does not become a fact because it has been repeated in several documents.

6. Communications log

Record the date, participants, subject, purpose, action required and link to the complete communication. Do not turn the log into a second inbox by copying every email into it.

7. AI-use and verification log

For any significant AI-assisted output, record:

• the date and tool used;
• the task requested;
• the source IDs supplied;
• whether identifying or sensitive information was removed;
• the output location;
• what was checked and against which source;
• errors found and corrected; and
• the final status: draft, checked or verified.

A folder and filename system that works

The folder structure should be understandable to someone who did not create it. Numbering keeps folders in a predictable order:

Keep especially sensitive information—such as confidential contact details, passwords or protected addresses—out of broadly shared folders. HMCTS guidance requires a C8 confidential contact details form sent by email to be a separate attachment and clearly named. Always follow the order and current filing guidance for your case.

A practical filename convention

Use a predictable format such as:

For example:

• 2026-08-12_ORDER_DJ-SMITH_DIRECTIONS_SEALED.pdf
• 2026-08-18_EMAIL_CAFCASS_APPOINTMENT_ORIGINAL.pdf
• 2026-08-24_POSITION-STATEMENT_APPLICANT_FILED.pdf

Use status labels consistently: ORIGINAL, WORKING, SIGNED, FILED, SERVED, SEALED or SUPERSEDED. Preserve the original electronic file and its metadata where they may matter. Editing a working copy must not alter or replace the source evidence.

How to connect the system to AI safely

“Connect” does not have to mean installing an automatic integration which continuously feeds the case into an AI model. In sensitive litigation, a deliberate manual hand-off is often the better control. Automation can spread an incorrect label, duplicate restricted information or act on the wrong version before anyone notices.

1. Preserve: save the complete original in the protected evidence store.
2. Index: assign a stable document ID and record its source, date, context and status.
3. Select: choose only the documents necessary for the defined AI task.
4. Minimise: remove names, addresses, children’s details, medical information and other data which the task does not require, where lawful and appropriate.
5. Constrain: tell the AI to use only the supplied documents, identify uncertainty and avoid inventing missing facts.
6. Cite: require a source ID and page for every material entry or proposition.
7. Compare: check the output line by line against the original sources.
8. Correct: amend the working draft and record any material AI error.
9. Finalise manually: place only the verified version in the filed-and-served area.

An example source-controlled prompt

The value of the prompt is not its length. It forces traceability. The human reviewer can move from each AI-generated entry back to the evidence rather than trusting a fluent narrative.

Keep fact, inference and law separate

An AI system may blur:

• what a person alleged;
• what a document proves;
• what someone inferred;
• what the court found; and
• what the law requires.

Your case hub should not. Those are different categories and should remain visibly different throughout the workflow.

The AI traffic-light test

The High Court’s decision in Ayinde v London Borough of Haringey; Al-Haroun v Qatar National Bank [2025] EWHC 1383 (Admin) illustrates the seriousness of false or unverifiable legal material reaching the court. The safest rule for a litigant is simple: never rely on an AI-generated case, quotation, statutory provision or procedural rule until it has been checked against an authoritative and current source.

JSH Law’s separate guides, The Legal AI Trust Test: Seven Checks Before You Rely on ChatGPT or Any AI in Court and How to Use ChatGPT as a Litigant in Person, explain these verification and project-organisation safeguards in more detail.

Confidentiality, privacy and family-court restrictions

A family-court file may include information about children, home addresses, schools, health, domestic abuse, criminal allegations, finances and third parties. Some information may also be subject to express orders, statutory restrictions or rules controlling communication outside the proceedings.

Family Procedure Rules 12.73 and 12.75 regulate communication of information in children proceedings held in private. The government’s guidance on sharing information outside family proceedings warns that documents and information cannot simply be shared with anyone, and that the court may make case-specific directions.

The judiciary’s October 2025 AI guidance is directed to judicial office holders rather than litigants, but its practical principles are instructive: protect confidential information, understand the risk of hallucination and bias, engage directly with the underlying evidence and retain personal responsibility.

The Information Commissioner’s Office guidance on AI security and data minimisation is primarily written for organisations, but minimising data and assessing security are sensible disciplines for anyone handling sensitive court information.

Before using a platform, check

• who can access the information;
• whether the material may be used to improve or train models;
• how long prompts, files and logs are retained;
• whether sharing links are public, restricted or capable of being forwarded;
• where account sessions and connected devices can be reviewed;
• whether files and records can be exported and deleted;
• whether deleted material remains in backups or logs;
• whether plugins, extensions or connected apps can see the data; and
• whether the account is personal, shared, managed by an employer or accessible to another household member.

Use data minimisation

If the task does not require a child’s name, address, date of birth, school, medical diagnosis or other identifying information, do not provide it. Replace names with stable labels such as Parent A, Parent B and Child 1, while keeping a private key outside the AI workspace where necessary.

Never record a hearing without permission

AI transcription does not create an exception to the rules on court recording. Family Procedure Rule 27.9 states that a party or member of the public may not use unofficial recording equipment without the court’s permission. Doing so without permission constitutes contempt of court under section 9 of the Contempt of Court Act 1981.

Why your case system is not your court bundle

Your private case system may catalogue everything potentially relevant. A court bundle is a selected procedural document prepared for a particular hearing. Confusing the two can produce an enormous, repetitive bundle which obscures rather than clarifies the case.

The revised Practice Direction 27A: Family Proceedings—Court Bundles came into force on 2 March 2026. The judiciary also published a dedicated guide for litigants in person.

What the current rules mean in practice

• The order and any case-specific direction come first.
• Subject to the rules and the court’s directions, the default position is that the applicant prepares, files and serves the bundle.
• If the applicant is a litigant in person and the respondent is represented, the represented respondent normally assumes responsibility, subject to any court direction.
• If all parties are litigants in person and the court is satisfied none can prepare the bundle, the court may direct HMCTS to do so. This is exceptional and not automatic.
• The bundle should contain only documents relevant to the hearing which the court needs to read or which will actually be referred to.
• Correspondence, emails, text or WhatsApp messages, recordings, financial records and photographs are not included merely because they exist; a specific item may be included where it is necessary for the court to read or will actually be referred to, or where the court directs.
• An e-bundle is ordinarily limited to 350 A4 pages unless the court permits otherwise.
• Significant documents and sections must be bookmarked; typed text in scanned pages must be made searchable by OCR; pagination and PDF numbering must comply with the applicable requirements.
• Preliminary documents should, so far as practicable, cross-refer to the relevant bundle pages.

This is exactly where a good case hub helps. It allows you to identify potentially relevant material without putting all of it before the judge. It can generate a proposed index, show duplication, locate the most recent version and connect chronology entries to the pages ultimately selected.

Emailing the family court

The current HMCTS email guidance for family proceedings contains detailed requirements. At the date of publication, it includes requirements about subject lines, separate attachments, file types, page limits and a 25MB total email limit. It also states that a C8 confidential contact details form must be a separate attachment and clearly named.

Your CRM should therefore record not only that something was “sent”, but the exact version, destination, date and time, method, recipients, attachment list, and delivery or filing evidence. Always recheck the live guidance before sending because administrative requirements can change.

For a broader document audit before seeking practical assistance, see JSH Law’s Family Court Case Audit: 12 Documents to Organise Before Asking for Help.

For practical assistance organising documents, building a source-linked chronology and preparing material for a hearing, see JSH Law’s evidence and chronology support. You can also book a consultation to discuss a defined scope of support.

A trauma-informed and abuse-aware system

Case management is not merely an efficiency exercise. Many litigants are trying to organise painful information while dealing with fear, sleep loss, financial pressure, coercive control or the repeated emotional impact of reading allegations and safeguarding material.

A humane system should reduce avoidable exposure and cognitive load:

• Use neutral folder and document names rather than emotionally charged labels.
• Keep a separate private reflections note; do not mix every feeling into the evidential chronology.
• Create a “not for today” or parking list so that non-urgent concerns do not derail deadline work.
• Use a dashboard to show only the next hearing, next deadlines and present priorities.
• Turn off unnecessary previews and notifications containing sensitive case information.
• Do not require yourself to reopen distressing evidence every time you need a date; use stable, accurate metadata and links.
• Record the source once and reuse the reference rather than repeatedly copying the underlying content.

Where technology may be monitored

If a former or current partner may have access to your device, email, shared cloud account, family subscription, location settings or password recovery method, creating a new digital case system can itself create risk. Use a safe device where possible. Do not make sudden account or security changes if doing so might alert an unsafe person or increase danger without first considering a safety plan.

Refuge Tech Safety provides specialist information about technology-facilitated abuse. The National Domestic Abuse Helpline’s technology-safety guidance advises anyone concerned that a device is being monitored to use a safe device. In an emergency, call 999.

A practical setup plan

Phase 1: secure the foundation

1. Choose a safe device and one primary account.
2. Use a passkey where available or a strong, unique password.
3. Turn on two-step verification and review recovery methods and logged-in devices.
4. Create the root case folder and separate original, working, filed and AI-material areas.
5. Arrange an independent, protected backup.

The National Cyber Security Centre recommends passkeys where available and otherwise strong unique passwords supported by two-step verification. See its guidance on securing important online accounts.

Phase 2: establish procedural control

1. Enter the case number, court, next hearing and latest order.
2. Extract every live direction into the orders and deadlines register.
3. Add calendar reminders well before each deadline.
4. Record the applications currently before the court and the decisions sought.
5. Identify any uncertainty requiring clarification or legal advice.

Phase 3: register the evidence

1. Preserve originals before renaming, converting, cropping or annotating anything.
2. Assign stable IDs to the important documents.
3. Build the chronology from source material, not memory alone.
4. Separate allegations, responses, evidence and findings.
5. Identify missing context, incomplete message threads and superseded versions.

Phase 4: introduce AI cautiously

1. Choose one narrow task.
2. Confirm that the selected material may lawfully and safely be processed.
3. Minimise identifying information and supply only the necessary working copies.
4. Require stable source references.
5. Verify the result and log material errors.
6. Keep final filing and service under deliberate human control.

Frequently asked questions

Do I need to buy a CRM to represent myself in family court?

No. For a single case, a secure folder structure, spreadsheet and calendar may provide all the CRM functions you need. Buy or adopt a more complex platform only if the case genuinely requires linked records or collaboration which the simpler system cannot manage safely.

Is a spreadsheet really enough?

Often, yes. A spreadsheet can maintain an orders register, chronology, evidence index, issues matrix, contacts list and communication log. Its limitations arise where relationships become complex, permissions are difficult to manage or multiple users overwrite information. Regular exports and backups remain important.

Can I upload my whole family-court bundle to ChatGPT or another AI service?

Do not assume that you can. Consider the court order, family-court communication restrictions, the nature of the documents, the purpose of the upload and the provider’s current data arrangements. Use the minimum material required and remove unnecessary identifying information. If the legal position is uncertain, obtain advice or seek a direction from the court before uploading.

Can AI write my witness statement?

AI may help organise headings or improve the clarity of a draft based on your own truthful account. It cannot supply your memory, decide what you witnessed or safely “improve” the facts. You must read every word, ensure it is true and accurate, and understand the significance of signing a statement of truth.

Does my CRM replace the court bundle?

No. The case hub is your comprehensive control system. The court bundle is a selected set of documents prepared for a particular hearing in accordance with the court’s order, PD27A and any applicable guidance.

Which technology ecosystem is best?

The best system is usually the simplest reputable ecosystem you already understand and can secure, back up and export. A Microsoft-centred, Google-centred or carefully chosen modular system can all work. Product choice matters less than disciplined source control, permissions, consistent records and verification.

Can I use an AI transcription app during a hearing?

Not without the court’s permission. Family Procedure Rule 27.9 prohibits unofficial recording equipment without permission, and unauthorised recording may constitute contempt of court. Ask about an official transcript or an appropriate direction rather than recording secretly.

Final thought: preparation is architecture

A family-court case is not strengthened by having the most documents or the most sophisticated software. It is strengthened when the relevant facts can be traced to reliable evidence, deadlines are controlled, versions are clear and the court can understand what decision is required.

The most important lesson from the Protiviti article is not that every organisation needs a newer CRM. It is that technology works only when the underlying operating model is coherent. The same is true for litigants in person.

Build the foundation first. Preserve the evidence. Create one reliable index. Control the deadlines. Use AI for carefully defined assistance. Check everything that matters.

The purpose of the system is not to automate responsibility. It is to make responsible participation possible.

Sources and further reading

• Sarah Riesenberg, Protiviti Technology Insights: Beyond CRM: Architecture Firm Discovers an Integrated Operating Model for AI-Driven Growth
• Sarah Riesenberg’s Protiviti author profile
• Adrian Anglin on LinkedIn
• Practice Direction 27A: Family Proceedings—Court Bundles
• Judiciary: Preparing Court Bundles for Family Proceedings—Guide for Litigants in Person
• HMCTS email guidance: family proceedings
• Family Procedure Rules Part 12: communication of information in children proceedings
• GOV.UK: Sharing information outside of court in family proceedings
• Family Procedure Rules Part 27: hearings, recordings and transcripts
• Judiciary: Artificial Intelligence Guidance, October 2025
• Ayinde v London Borough of Haringey; Al-Haroun v Qatar National Bank [2025] EWHC 1383 (Admin)
• Information Commissioner’s Office: AI security and data minimisation
• National Cyber Security Centre: Secure your important online accounts
• Refuge Tech Safety

Sources and procedural links checked on 30 August 2026. Court rules, platform settings and administrative requirements can change. Always check the current version and the directions made in your own case.

About the author

Jessica Susan Hill is the founder of JSH Law. She supports litigants in person with practical family-court preparation, document and evidence organisation, chronologies, statements, procedural mapping, hearing support and responsible legal-technology workflows. She writes about access to justice, family justice, domestic abuse, safeguarding and the responsible use of artificial intelligence in legal systems.