Business AI is already changing how lawyers work - ready or not
Lawyers aren't replacing themselves with AI. They're using it to shave hours off low-value work so they can spend more time on strategy, judgment, and client care. Tools like ChatGPT, Claude, Copilot, Gemini, and Grok are now part of the legal toolkit - with guardrails.
The pattern is clear: use AI as a first pass, keep humans as the final authority.
Where firms are getting results now
Discovery and evidence review. Derek Martin, co-founder at Driver Defense Team (Stone Park), said his office leans on AI to transcribe and organize video evidence from police body cams and dash cams. "It helps to summarize timelines and identify key moments for attorney review and allows our lawyers to spend less time searching and more time thinking about case facts and strategy." For teams building similar workflows, start with dependable transcription and summarization pipelines. See: Speech-To-Text.
Research and document search. Securities investment fraud attorney Andrew Stoltmann (Hoffman Estates) uses an AI assistant to find patterns, trends, and key documents. "AI has probably cut down legal research time from 20 hours a month down to about one hour a month. It has been nothing short of amazing."
Client intake and communications. Divorce attorney Raiford Dalton Palmer (Naperville) is implementing Lexidesk to handle new client calls and texts when staff are unavailable. His firm also uses ChatGPT Teams to share research, Harvey.ai for drafting and analysis, and is piloting Lexis Protégé for guided workflows and secure automation. "The tools are developing so quickly that it is difficult to stay on top of all the new offerings and to know which are best for our purposes."
The risk side - and how top attorneys are handling it
Media stories about hallucinated citations and sanctions have put everyone on alert. Keith Shindler (Schaumburg) summed it up: "I've read about lawyers, even some judges who come under scrutiny for possibly using AI and not sufficiently investigating the results of the AI prompt - and citing those in legal documents without doing the proper verification. That's not anything I want to be mixed up in."
He verifies everything before citing it, with special attention to case law and client data. That's the standard to adopt.
George Bellas, partner at Bellas & Wachowski (Park Ridge), first chair of the Illinois State Bar Association's AI Committee and former member of the Illinois Supreme Court's Commission on AI, uses AI for research, drafting memos, brainstorming options, deposition prep, document summaries, and timelines. His warning is worth repeating: "My concern is that it will stifle creativity in lawyers and they will rely on the stock answers provided by AI agents. The profession will contract as lawyers employ AI in their practices and many jobs will be lost. There are many new ethical issues which are coming to the surface and the profession must adapt to the issues facing us in the use of AI."
Practical guardrails you can adopt this quarter
- Verification workflow: Require citations and quotes from source documents. Shepardize/KeyCite every case before it hits a draft. Use a second model or human peer check for any legal conclusion.
- Data handling: Keep client data in enterprise tools with contractual privacy. Strip PII before prompts. Lock down uploads and export logs.
- Policy statement: "AI is a first-pass assistant, not a final authority." Derek Martin put it plainly: make AI part of a structured, multipass review that preserves human judgment where it matters most.
- Ethics alignment: Map usage to tech competence duties under ABA Model Rule 1.1. Use a risk checklist aligned to the NIST AI Risk Management Framework.
- Audit trail: Save prompts, versions, and source packs in the matter file. Make "show your work" non-negotiable.
- Team training: Set standards for prompts, privacy, and verification. For structured learning, see AI for Legal.
Workflow templates you can copy
- Discovery acceleration: Auto-transcribe audio/video, split by speaker, summarize by issue, extract entities (people, places, timestamps), then generate a draft timeline with links back to source clips for fast verification.
- Research sprint: Start with an AI outline of issues and potential authorities. Pull full texts from trusted databases. Verify every cite. Have AI generate counterarguments to stress-test your position.
- Client intake triage: AI answers FAQs, captures facts in a structured form, conflicts-checks keywords, and schedules a consult. Human reviews the intake packet before outreach.
- Deposition prep: Feed key documents and prior testimony to generate draft outlines, topic trees, and follow-up questions. Attorney edits, adds exhibits, and locks final sequencing.
Tooling snapshot from firms in the field
- Core assistants: ChatGPT Teams, Claude, Copilot, Gemini, Grok
- Legal-specific: Harvey.ai (drafting, research, review), Lexis Protégé (guided workflows, automation), Lexidesk (intake via calls/texts)
- Tip: Evaluate security posture, data retention, and audit features before any rollout. Pilot with non-sensitive matters first.
Metrics to watch
- Hours saved per matter (research, drafting, review)
- First-response time on new intake (off-hours included)
- Verified citations per memo and correction rate on AI drafts
- Cost per matter vs. baseline and client satisfaction scores
- Share of tasks completed by AI as a first pass (with human sign-off)
Bottom line
Adopt AI where it trims hours and improves client access. Keep your standards high: verify, protect data, and document your reasoning. Treat AI as the first draft assistant - never the final decision-maker.
Your membership also unlocks:
