AI Intake and Law Firm Conflict Checks: What to Automate and What Lawyers Must Review

By Attorney Michael Brunman, Co-Founder and CEO of Clerx

A conflict-aware intake workflow is a process that collects only the information needed for a preliminary conflicts search, checks that the names are complete and usable, routes the record to the firm's approved system, and prevents the conversation from advancing until an authorized person clears the next step. Automation can support this process. Lawyers remain responsible for legal analysis, professional duties, waivers, screening, and acceptance of representation.

This distinction is important because the first inquiry is not ethically empty space. Under ABA Model Rule 1.18 and applicable state rules, duties can arise when a person consults a lawyer about a possible client-lawyer relationship. A law firm should design intake so it is responsive without encouraging unnecessary disclosure before the firm knows whether it can proceed.

Conflict collection is not conflict clearance

Three activities are often compressed into the phrase “conflict check”:

1. collecting names and matter identifiers;
2. searching records for possible matches;
3. deciding whether a conflict exists and what the firm may do.

The first two may include automation. The third requires professional judgment under the firm's policies and governing rules.

A Clerx AI receptionist can ask firm-approved questions, confirm spelling, distinguish people from organizations, and create a structured handoff. The firm's practice-management or conflicts system may search existing records. An attorney or authorized conflicts professional then reviews the result and decides whether the intake may continue.

The workflow should never convert “no obvious database match” into “no conflict.” Records can be incomplete, names can vary, related parties may be missing, and legal significance depends on facts a search result cannot resolve.

Why intake design affects conflicts risk

An eager intake process may ask for a complete story before collecting the names needed for screening. That creates two operational problems. First, sensitive information spreads across transcripts, emails, texts, notes, and notifications before the firm decides whether it can act. Second, staff may invest time in a detailed intake that should have paused earlier.

The safer design is staged disclosure:

• explain that the interaction is preliminary;
• collect identity and contact information;
• identify the broad matter category;
• collect the minimum involved names required by the firm;
• run the approved review step;
• continue detailed intake only after the next gate is cleared.

The law-firm intake form guide applies the same principle: ask information when it is needed, not merely because the form can hold it.

What names should the workflow collect?

There is no universal list. The required names depend on practice area, jurisdiction, firm policy, and matter type. Firms should define a minimum dataset for each intake branch.

Examples may include:

• the prospective client's full legal name and prior names;
• opposing or adverse parties;
• spouses, former spouses, children, or household members in family matters;
• co-defendants, alleged victims, witnesses, or agencies in criminal matters when the firm requires them;
• employers, petitioners, beneficiaries, or sponsored relatives in immigration matters;
• deceased persons, fiduciaries, beneficiaries, family members, and entities in trusts and estates;
• companies, affiliates, owners, insurers, and key individuals in business matters;
• related cases, courts, case numbers, or transaction names when relevant.

The intake should confirm spelling and role. “John Smith” is less useful than “John A. Smith, prospective client's former spouse.” It should also allow uncertainty. A caller may not know a legal entity's exact name, and the workflow should flag that fact rather than invent precision.

A seven-stage conflict-aware intake workflow

1. Set expectations before collecting details

Use approved language explaining that the firm is gathering preliminary information, has not agreed to represent the person, and cannot provide legal advice through the intake channel. Encourage the person not to share a detailed confidential narrative until asked.

This message should be understandable, not a wall of legal text. The aim is to create a clear boundary while preserving a respectful client experience.

2. Classify the person and request

Determine whether the contact is a new prospective client, current client, former client, opposing party, witness, vendor, or referral source. The new-lead and existing-client routing framework prevents a current client from entering a new-prospect conflicts path and helps unexpected callers reach the appropriate process.

Identify the broad practice area and matter category without inviting the full story. This tells the workflow which names and fields are required.

3. Collect and normalize involved names

Ask the minimum firm-approved questions. Confirm spelling aloud on calls when practical. Separate first, middle, and last names. Capture aliases or prior names in dedicated fields. Distinguish a person from a company and record the person's role.

On law-firm text messaging, avoid placing sensitive allegations in the first message. A shared phone or lock-screen preview may expose the subject of the inquiry. Use text to verify identity and gather limited routing information, then move detailed facts into the firm's approved process.

4. Create the preliminary search record

Send the structured record to the firm's approved destination. The Clerx integrations directory includes practice-management and CRM connections, but each firm should verify exactly which objects, fields, permissions, and notifications are used.

Deduplicate before creating another lead. A returning prospect, former client, current client's relative, or person who used several channels may already exist.

5. Pause at the decision gate

Do not automatically offer a substantive consultation merely because intake is complete. The workflow should create a review task for an authorized person, state the response target, and preserve a neutral status such as “pending conflicts review.”

If the firm permits certain low-risk pathways to book before final clearance, document exactly what the appointment means and which information may be collected. The booking itself should not imply representation.

6. Route the outcome without overexplaining

If cleared for the next step, continue qualification and offer the correct consultation. The firm's consultation-booking workflow should define the calendar, fee, confirmation, and handoff.

If the firm cannot proceed, use approved language. Intake staff and automation should not speculate about which client, matter, or relationship caused the result. The response may simply state that the firm cannot assist and, if policy permits, provide neutral next-step information.

7. Retain, restrict, and review the record

Define who can access preliminary intake records, how transcripts and recordings are handled, how long unretained inquiries are kept, and how information is searched in future checks. The AI receptionist security checklist provides a broader vendor and data review framework.

The firm should also decide how to correct misspelled names, merge duplicates, and document the person who cleared the step.

What AI can safely automate

Within a firm-approved workflow, automation can support:

• immediate acknowledgment;
• caller classification;
• collection and confirmation of names;
• standard formatting and role labels;
• duplicate detection signals;
• creation of a pending-review task;
• routing to a primary and backup reviewer;
• reminders when the review target is missed;
• continuation to approved questions after clearance;
• consistent disposition and audit fields.

These are operational tasks. They improve speed and consistency without asking the system to interpret ethics rules.

What should remain with lawyers and authorized staff

Human review should own:

• whether the person is a prospective client under applicable law;
• whether a conflict exists;
• whether matters are the same or substantially related;
• whether information could be significantly harmful;
• whether consent or screening is available and sufficient;
• whether the firm may accept, decline, or refer the matter;
• what the firm may communicate about the decision;
• any waiver, ethical wall, or jurisdiction-specific requirement.

An AI receptionist versus intake specialist framework helps teams separate repeatable operational steps from judgment-heavy decisions.

Practice-area examples

Family law

Collect the prospect's name, spouse or former spouse, children when required, related household members, and known court information. Pause before asking for a detailed account of abuse, finances, or strategy unless the firm has approved that stage.

Criminal defense

Collect the prospective client's name, jurisdiction, broad charge or event category, court or agency, and other names the firm requires. The system may recognize a stated arrest or near-term court date for priority routing, but it should not judge legal urgency or discuss defenses.

Immigration

Collect the prospective client's name, petitioner or sponsor, employer when relevant, family members involved, and broad matter type. Do not treat the existence of a government deadline as proof of legal urgency. Route the stated date for attorney review.

Estate planning and administration

Distinguish the person seeking advice from a deceased person, spouse, fiduciary, beneficiary, or family member. Collect limited role information before receiving a full description of family conflict.

Test the process with adversarial scenarios

A normal test call is not enough. Test:

1. a common name with several database matches;
2. a misspelled adverse-party name;
3. a person using a former name;
4. a caller who refuses to identify the opposing party;
5. an opposing party attempting to provide extensive information;
6. a current client calling through a lead form;
7. one prospect contacting the firm by call, chat, and SMS;
8. a professional referral with incomplete names;
9. a reviewer who does not respond;
10. a qualified prospect who tries to book before clearance;
11. a caller requesting legal advice during the pause;
12. a system outage or failed integration.

Use the firm's intake SLA to assign the review owner, backup, response target, and escalation path.

Measure quality, not only speed

Useful metrics include name-completion rate, duplicate rate, percentage of records requiring correction, time to conflicts review, percentage breaching the target, prospects lost during the pause, unauthorized bookings, access exceptions, and conversion after clearance.

Fast intake is valuable only when the record supports the next decision. The intake dashboard metrics should therefore include data quality and handoff reliability alongside response and booking rates.

Frequently asked questions

Can AI perform a law firm conflict check?

AI can collect and normalize names, support database searches, and route a record for review. It should not make the final legal conflicts determination.

What is the difference between conflict collection and clearance?

Collection gathers the names and matter identifiers needed for a search. Clearance applies legal and firm judgment to the results and decides whether intake may proceed.

When should detailed intake begin?

Use staged disclosure. Gather the minimum needed for the firm's preliminary review, then continue detailed intake after the appropriate gate is cleared.

Should a prospect be allowed to book before clearance?

That depends on firm policy and the nature of the appointment. If allowed, the workflow should state that booking does not create representation and should limit information collection accordingly.

What if the caller will not provide an adverse party's name?

Follow the firm's approved exception path. The system should not invent information or bypass the review requirement.

Can automation tell a person why the firm declined?

Use firm-approved language and avoid revealing another client's identity or confidential information. A neutral statement is often safer than speculation.

How should duplicate inquiries be handled?

Match verified identity and contact information, merge or link records under firm rules, and preserve the history across channels.

Does a disclaimer eliminate prospective-client duties?

No single disclaimer resolves every issue. The workflow, the person's reasonable expectations, the information exchanged, and applicable rules all matter.

What should firms test before launch?

Test misspellings, aliases, common names, opposing parties, duplicates, early booking attempts, reviewer delays, legal-advice requests, and system failures.

What should remain under attorney control?

Legal conflicts analysis, ethical decisions, waivers, screens, matter acceptance, and communications requiring professional judgment.

Book a demo with Clerx today.