More Maryland law firms adopt AI as the tech gets practical
Key takeaways
- Maryland firms are increasing AI use for discovery and case analysis.
- ABA data shows nationwide law firm adoption grew from 11% to 30% between 2023 and 2024 (published in 2025).
- Common uses: document review, issue summaries, and drafting pleadings or openings.
- Training, verification, and oversight are essential to avoid errors and hallucinations.
Momentum backed by data
Law firms across Maryland are putting AI to work in day-to-day matters. That tracks with an American Bar Association survey showing adoption among firms jumped from 11% to 30% between 2023 and 2024, with results released in 2025. ABA tech reports have been signaling this direction for years.
"Although AI has been around, it has exploded in the last year," said Mary Roby Sanders, a partner at Turnbull, Nicholson & Sanders in Towson.
Discovery: fast triage, smarter review
Sanders' team uses AI to grind through financial records in discovery. "We can scan in all kinds of bank records and credit card statements, and it will analyze those and give us a report," she said. Need every charge tied to restaurants or hotels? Give the instruction and let the tool pull it together.
It's also useful for spotting whether records were changed. If a client used QuickBooks, for example, the system can flag entries that look updated or out of place. "It takes an enormous amount of time to (manually) sort through these records and find the needle in the haystack, while AI can find it right away," Sanders said.
Case ramp-up and first drafts
For issue summaries, the goal is speed with traceability. "If we've had cases that are coming from other firms that have already had extensive litigation or discovery, we can put in the discovery that's been produced in the past for an issue summary," said Faith Khan, an associate at the firm. "It helps familiarize us with the case a lot quicker."
Source control matters. "The good thing about the program we use is it will bookmark every document that it pulls information from so we can verify that the information is accurate," Khan said. Her team also prompts AI for opening statement drafts and witness question ideas to jump-start preparation.
Smarter use and more models
Lawyers are getting sharper about what AI is good at-and where it can go off the rails. "I think generally speaking people are more aware of issues like hallucinations now than they were a year ago," said Matthew Kohel, a partner at Saul Ewing in Baltimore who chaired the MSBA task force that issued a generative AI advisory.
Firms are also mixing tools. "You might like how this LLM writes, while (another) one has better graphics or better research," Kohel said. Expect continued improvement across IP searches, contract drafting, and reviewing pleadings.
Training is the difference
"There's a whole skill set about being able to use the technology and being able to understand how it works, understand what you need to put into it, how you need to prompt it, and how you need to ask it to produce the results you're looking for," said Jason Balog, a principal at Miles & Stockbridge in Baltimore. He noted how quickly quality has improved over the past few years.
Firms have taken two paths: let people figure it out, or build training programs. Miles & Stockbridge chose structured training, and Balog credits it with better outcomes and safer use.
Practical checklist for your firm
- Define clear use cases: discovery triage, timeline building, issue spotting, first-draft summaries.
- Set verification rules: require citations, links, or document bookmarks; "no source, no use."
- Protect confidentiality: keep client data out of public tools; use enterprise controls and redaction.
- Create prompt standards: jurisdiction, facts, tone, and output format; include exclusions and accuracy checks.
- Route by task: choose the model that writes best for briefs vs. the one that analyzes data most reliably.
- QC every output: human review, cite-checks, and a second set of eyes for anything filed or sent to clients.
- Track outcomes: time saved, cost per matter, and error rates to refine workflows and justify licenses.
- Train continuously: short onboarding, office hours, and a living playbook with examples and red flags.
If your team needs a structured path to prompt skills and role-specific workflows, explore curated AI training by job role.
The bottom line: AI is now a practical tool in Maryland practices-especially for discovery, analysis, and first drafts. With the right training, guardrails, and verification, it saves hours without risking accuracy or confidentiality.
Your membership also unlocks:
