A tenant uses an online self-help tool after receiving court papers. The questions are clear. The tool identifies a possible defence, produces a neat summary and ends with the address of a legal-aid office. The tenant has done everything the product asked, yet still has no appointment, no assurance that the office can take the case and no indication that tomorrow’s deadline has reached anyone who can act.
On the product dashboard, this may count as a successful completion. In the person’s life, it may be a failed legal journey.
The handoff is the legal-tech product. A self-help service should be judged primarily by whether it transfers a person, with informed consent, the necessary context and any deadline, to a professional or institution able to take the next step. Information and document assembly can be valuable, but they are components of that transfer. When a matter requires judgment, representation, mediation or navigation, a link to a directory is not access to justice.
From a referral to a transfer of responsibility
A cold referral gives the user somewhere else to try. It may display a telephone number, produce a list of providers or instruct the person to begin another application. Each item can be accurate while the pathway still fails. The organisation may serve a different area, have no capacity, exclude the relevant subject, require documents the user does not have or discover a conflict only after the person has retold a sensitive story.
A warm handoff has a different mechanism. The system first identifies the kind of help required: legal advice, legal-aid representation, mediation, a court navigator or another social service. It checks current eligibility and availability at a level appropriate to the risk. It asks the person whether specific information may be sent. The receiving service acknowledges the referral, preserves the deadline and either offers the next contact or returns the matter through a defined fallback route. The person can see what was transferred and who is expected to respond.
This is increasingly part of the official conception of people-centred justice. The OECD’s 2025 implementation toolkit for access to justice calls for “no wrong door” entry, common intake, interoperable referrals and warm handoffs so people can move from information to advice and resolution without repeatedly telling their story. It also recommends monitoring handoffs, time to resolution, outcomes and drop-offs. The unit of success is the pathway, not the screen.
The receiving human matters more than the routing label
Legal problems rarely announce the professional category they require. A debt dispute may also involve housing instability. A family matter may contain an immediate safety risk. A benefits problem may be an administrative correction, an appeal or a missed evidence request. Classification software can organise these signals, but it should not silently turn an uncertain account into a definitive legal diagnosis.
The handoff therefore needs a named function at the other end. A court navigator can explain procedure and help someone find the right counter, but cannot supply legal strategy where local rules reserve that work to lawyers. A mediator can support negotiation where participation is safe and appropriate, but does not replace independent advice. A legal-aid worker can assess eligibility and urgency, while a lawyer can advise on choices and assume professional duties. Technology improves the pathway when it brings the person to the right competence; it becomes hazardous when a generic “help” destination conceals what the recipient can and cannot do.
The current legal-aid checker for England and Wales illustrates the distinction. The GOV.UK service asks about the problem and financial circumstances, then directs the user towards a telephone adviser, a local legal adviser or a mediator. It explicitly says that the relevant professional determines whether legal aid is available and that a legal adviser, not the applicant alone, makes the application. The digital step supports entry; it does not pretend to complete the professional decision.
Consent must attach to the information, not the button
A smooth transfer can tempt designers to collect a complete narrative once and distribute it widely. That is operationally convenient and legally dangerous. An intake account may contain allegations, health data, immigration status, children’s information, addresses or the identity of an opposing party. The user may reasonably think they are speaking within a protected legal relationship when they are actually submitting data to a platform, contractor or referral network.
Confidentiality, legal professional privilege and data protection are related but not interchangeable. Their scope depends on the jurisdiction, the recipient, the purpose and whether a professional relationship has formed. A self-help interface should never give a blanket assurance that everything entered is privileged. It should say who operates the service, who will receive the material, for what purpose, and which protections apply before asking for sensitive facts.
The safer design is progressive disclosure. Routing can often begin with a location, broad problem type, urgency, preferred contact channel and the names needed for a conflict check. A fuller narrative should follow only when a suitable recipient has been identified and the person has agreed to the transfer. This is consistent with the GDPR principles of purpose limitation and data minimisation: personal data should be collected for specified purposes and be adequate, relevant and limited to what is necessary.
Conflict checks belong before the story
A legal provider may be unable to act because it represents, or has relevant confidential information about, another party. If the platform sends the whole account before checking, it can expose details unnecessarily and may complicate the recipient’s ability to accept the matter.
Guidance in England and Wales makes this sequencing concrete. The Solicitors Regulation Authority advises firms to limit confidential information obtained before a conflict check and says client information should not be passed to a third party by way of referral without consent. Its current Code of Conduct also requires solicitors to protect current and former clients’ affairs and not to act where specified conflicts or risks arise, subject to defined conditions.
The exact rules differ elsewhere, but the product lesson travels well. Separate the minimum identity and adverse-party data needed to test whether a recipient may engage from the facts needed to advise. Do not make the person disclose the most sensitive part of the case to discover that the door was closed all along.
The strongest case for self-service
There is a serious counterargument. Legal assistance is scarce. Human intake is expensive. Automated tools can operate outside office hours, serve people at a distance and standardise questions that busy staff may otherwise omit. Requiring a personal handoff for every user could consume professional capacity on matters that some people can resolve with clear information and a form.
That argument is right about the value of self-service and wrong if it treats self-service as the default destination. Some people want information and can act on it. Their independence should not be obstructed. The product should let them complete an appropriate pathway without unnecessary professional intervention. But the system must detect when the route changes: a deadline, vulnerability, unclear facts, disputed identity, power imbalance or need for individual legal judgment should open a human channel. The choice is not automation or people. It is automation that preserves a reachable human boundary.
Nor can software create capacity that does not exist. A beautifully orchestrated handoff fails if legal-aid offices have closed their lists or navigators are overloaded. This is not a reason to revert to a directory. It is a reason to make the shortage visible. Live capacity data, rejected-referral reasons and waiting times expose where public funding, staffing or service redesign is required instead of converting scarcity into user error.
Measure arrival, not departure
Completion rates, generated forms and clicks on referral links describe what happened inside the tool. They do not show whether help arrived. Better measures follow the handoff: whether the recipient accepted it, how long first contact took, whether the deadline survived, whether the user had to repeat the account, whether the referral was redirected, and whether the person received the service the pathway promised. Outcomes should include practical changes in the person’s situation, not only case categories closed by institutions.
Outcome measurement must remain careful. A legal-aid service cannot guarantee a favourable judgment, and difficult cases should not make a provider look worse simply because it accepts them. The useful question is whether the handoff delivered timely, appropriate assistance and increased the person’s ability to make and act on informed choices. The Legal Services Corporation’s outcomes toolkit case study shows why follow-up matters: extended contact with recipients of limited legal services identified meaningful outcomes that conventional case-closing measures had missed.
Most importantly, the person should remain an agent in the transfer. They should be able to review the case summary, correct it, choose the contact method and decline data sharing without losing access to basic information. A human recipient should be able to question the intake rather than inherit its classifications as facts.
If you want to follow this series on technology that strengthens legal capability without displacing human responsibility, you can subscribe to Alkemata.
The remaining decision is institutional, not cosmetic: who accepts responsibility when the self-help tool recognises that self-help is no longer enough? Until a service can answer that question with a person, a deadline and a confirmed next step, its legal-tech product is unfinished.