1. Structure Intake Without Collecting the Whole Case Before Conflict Review
Good first-stage intake collects name, contact details, general matter type, opposing or related parties needed for conflicts, jurisdiction or location, any critical deadline if known, and a high-level summary. AI can summarize, classify, identify missing basic fields, route, and schedule. A chatbot should not encourage prospective clients to dump highly sensitive facts before the firm has designed its conflict process, confidentiality handling, retention, disclaimers, and handoff. Do not let AI decide the firm can represent someone.
AI can help
- ✓Summarize and classify prospective-client intake
- ✓Identify missing basic fields and route the inquiry
- ✓Schedule a consultation and send confirmation
Human / specialist keeps
- ●Decide whether the firm can represent the client
- ●Run the formal conflict and eligibility process
- ●Own confidentiality and retention decisions
Warning
Do not let AI decide 'we can represent you.' Representation is a human, conflict-checked, engagement-governed decision.
2. AI Can Prepare Conflict Information—It Should Not Replace the Conflict Database
AI may extract the prospective client, adverse parties, related entities, witnesses or organizations, and matter description from intake. Then the firm's conflict system or practice database performs the actual conflict process, and a human reviews the result. Do not rely on an LLM's memory to conclude that no conflict exists.
AI can help
- ✓Extract parties, adverse parties, and related entities from intake
- ✓Organize conflict-input information for review
- ✓Flag missing parties before the conflict check
Human / specialist keeps
- ●Run the formal conflict check in the practice database
- ●Review conflict results and decide on engagement
- ●Own the representation decision
Warning
Do not rely on an LLM's memory to conclude 'no conflict exists.' Use the firm's conflict system.
3. Use Voice AI for Administrative Intake, Not Unsupervised Legal Advice
Good Voice AI jobs include hours, locations, practice areas, basic process FAQs, contact capture, high-level matter category, consultation scheduling, transfers, task creation, and confirmation. It must not interpret law for the caller, predict outcome, promise representation, quote unusual fees unless approved, tell someone a deadline is safe, or give emergency legal instruction outside approved policy. Human handoff must be obvious.
AI can help
- ✓Answer hours, locations, and practice-area FAQs
- ✓Capture contact details and matter category
- ✓Book a consultation and transfer or create a task
Human / specialist keeps
- ●Provide legal advice and outcome predictions
- ●Promise representation and approve fees
- ●Handle emergency or sensitive legal instruction
4. Use AI to Accelerate Research—Never Skip Authority Verification
AI can help identify issues, discover search terms, summarize cases, compare authorities, build research plans, find contrary arguments, organize citations, and draft research memos. But an AI-generated citation is not verified authority. For legal conclusions, inspect the case, statute, rule, regulation, date, jurisdiction, precedential status, negative treatment, and subsequent history. Specialist legal AI can have an advantage when grounded directly in authoritative sources and citators.
AI can help
- ✓Identify issues and discover search terms
- ✓Summarize and compare authorities
- ✓Draft research memos and organize citations
Human / specialist keeps
- ●Verify every citation against authoritative sources
- ●Check jurisdiction, precedential status, and negative treatment
- ●Own the legal conclusion
Warning
An AI-generated citation is not verified authority. A plausible citation can be fake. Inspect the source.
5. Use AI to Review Documents Faster—Define the Job Precisely
AI can summarize an agreement, extract clauses, compare versions, identify defined terms, create a chronology, classify documents, surface inconsistencies, and prepare an issue list. Avoid vague prompts like 'tell me if this contract is safe.' Better: extract termination provisions, identify governing-law clauses, compare indemnification against an approved playbook, and list provisions requiring lawyer review. The human lawyer decides legal significance.
AI can help
- ✓Summarize agreements and extract clauses
- ✓Compare versions and surface inconsistencies
- ✓Prepare issue lists and chronologies
Human / specialist keeps
- ●Decide legal significance and risk
- ●Advise on provisions requiring negotiation or change
- ●Own the legal opinion
6. Draft From Verified Facts, Authority, and Firm Templates
Combine matter facts, verified research, firm template or precedent, and lawyer instructions into an AI draft, then run fact review, authority review, privilege and confidentiality review, and lawyer edit before finalizing. AI may draft client letters, memos, motion sections, clauses, demand letters, chronologies, discovery requests, and internal outlines. Do not file, send, or advise from unreviewed AI work.
AI can help
- ✓Draft first versions from verified facts and templates
- ✓Generate clauses, letters, memos, and outlines
- ✓Structure discovery requests and chronologies
Human / specialist keeps
- ●Review facts, authority, and privilege before finalizing
- ●Edit and approve all work product
- ●Own what is filed, sent, or advised
Note
Do not file, send, or advise from unreviewed AI work. Treat AI output as draft requiring lawyer review.
7. Use AI to Organize the Record—Keep the Original Source
AI may create a summary, timeline, key admissions, open questions, action items, and an issue map. Preserve the original recording, transcript, or document where law and policy permit. Do not treat the AI summary as the authoritative record. Recording and consent rules vary by jurisdiction.
AI can help
- ✓Create summaries, timelines, and issue maps
- ✓Extract key admissions and action items
- ✓Organize long records for review
Human / specialist keeps
- ●Preserve and rely on the original source record
- ●Manage recording and consent compliance
- ●Own the authoritative record
8. Draft Routine Communication Without Automating the Attorney-Client Relationship
Good AI jobs include appointment confirmation, document requests, status-update drafts, meeting recaps, approved process explanations, invoice reminders, and next-step checklists. Higher-risk work includes legal advice, settlement recommendations, litigation strategy, contract interpretation, deadline advice, and outcome prediction. Human review should increase with legal significance.
AI can help
- ✓Draft confirmations, reminders, and status updates
- ✓Prepare meeting recaps and document requests
- ✓Generate approved process explanations
Human / specialist keeps
- ●Provide legal advice and strategy
- ●Handle settlement, interpretation, and deadline advice
- ●Own the attorney-client relationship