AI Meeting Summaries That Save 1,300 Attorney Hours Per Year
AI-powered automation system that transforms Microsoft Teams meetings into structured summaries delivered via email, saving 1,300 hours annually for Minsky, McCormick & Hallagan P.C.

The Client
Minsky, McCormick & Hallagan P.C. (MMH PC) is a well-established Chicago immigration law firm founded in 1974. With over 50 years of experience and 20+ attorneys, the firm specializes in business immigration (H-1B, L-1, EB-1, EB-2, EB-3), family-based immigration petitions, and citizenship matters for clients across the United States.
Immigration law is documentation-intensive and detail-critical. Cases often span multiple years with dozens of client meetings covering fee agreements, required documents, case timelines, USCIS filing deadlines, and next steps. Attorneys must maintain precise records of every discussion—what documents were promised, what deadlines were communicated, what legal advice was given. This documentation serves both client service and legal liability protection.
The firm conducts 500+ client meetings annually across their 20-attorney team. Each meeting generates critical information that must be captured, communicated to clients, and archived for future reference. Many clients are non-native English speakers navigating complex US immigration processes, making clear written communication especially important. With multi-year cases that may transfer between attorneys, the firm needed comprehensive meeting documentation to ensure continuity of service.
The Challenge
After every client meeting, MMH PC attorneys faced the same tedious task: writing notes from memory. Each attorney would spend 15 minutes manually typing up what was discussed—fee agreements, required documents, case strategy, action items, and client questions. Despite this effort, the notes were always incomplete. Human memory is imperfect, especially after back-to-back meetings covering different cases with similar issues.
The math was brutal. With 20 attorneys each conducting approximately 5 client meetings per week, the firm held about 500 meetings annually. At 15 minutes of post-meeting note-taking per meeting, that was 1,300 hours per year spent on manual documentation—time that could have been spent on billable legal work or business development.
But the bigger problem wasn’t just time—it was accuracy and communication. Clients received no written confirmation of meeting discussions. Days later, they’d email asking “What documents did you say I need?” or “When did you say my filing deadline is?” Attorneys had to search through incomplete notes or try to remember details from a meeting held weeks ago. For non-native English speakers, verbal instructions without written confirmation led to misunderstandings and missed deadlines.
Multi-year immigration cases create another challenge: meeting history. An H-1B case might involve 10-15 meetings over 3 years discussing different phases (initial filing, RFE response, extension, green card transition). When a case transferred between attorneys—due to vacation, caseload balancing, or attorney departure—the new attorney had to piece together context from fragmented manual notes scattered across different systems. There was no searchable, comprehensive meeting archive.
The documentation gaps created malpractice risk. If a client claimed “You told me the filing deadline was October 15th” but the attorney’s notes said October 1st (or worse, had no deadline recorded), the firm had no objective record of what was actually discussed. The firm needed automated meeting documentation that was comprehensive, searchable, delivered instantly to clients, and legally defensible.
Our Solution
We built an AI-powered automation system that transforms Microsoft Teams meetings into structured summaries automatically delivered via email within minutes of meeting completion. The system requires zero manual work from attorneys—just hold the meeting with transcription enabled, and everything else happens automatically.
The architecture uses Microsoft Graph API to monitor the firm’s Microsoft 365 tenant for completed Teams meetings. We implemented a triple-redundant discovery system that checks for meetings in three ways (calendar events, user activity, Teams endpoint polling) to ensure no meeting is ever missed. When a meeting completes, the system downloads the auto-generated transcript in WebVTT format.
The transcript is processed by Azure OpenAI Service using GPT-4, which has been custom-prompted to generate structured summaries in a legal services format. The AI extracts key discussion topics, action items with responsible parties, client questions and attorney answers, fee agreements or payment discussions, required documents, filing deadlines, and next steps. The summary is formatted in clear sections that match how attorneys think about case documentation.
The structured summary is automatically emailed to all meeting participants via Resend API. Attorneys receive their meeting notes instantly without lifting a finger. Clients receive written confirmation of everything discussed—no more “what did we talk about?” follow-up emails. All emails are threaded by meeting subject line, creating a searchable archive that both attorneys and clients can reference months later.
The system includes intelligent deduplication (some meetings generate multiple transcript versions as Microsoft processes them), error handling for network issues or API rate limits, and fallback mechanisms if any step fails. OAuth 2.0 authentication ensures secure access to the firm’s Microsoft 365 data. The entire system runs on Azure Functions, scaling automatically to handle multiple concurrent meetings without manual intervention.
The Impact
The Teams Meeting Summarizer eliminated 1,300 hours of annual manual note-taking work—time that attorneys can now spend on billable legal work, client development, or professional development. The 15-minute post-meeting note-writing ritual is gone, representing a 99% time reduction in meeting documentation.
More importantly, meeting documentation quality improved dramatically. AI-generated summaries capture everything discussed with perfect recall—no details forgotten, no action items missed. Attorneys can focus entirely on the client conversation during meetings instead of mentally preparing what to write later. The summaries often include discussion points that attorneys would have forgotten to document manually.
Clients now receive written confirmation of every meeting within minutes of completion. This dramatically reduced “what did we discuss?” follow-up emails and improved client satisfaction. Non-native English speakers particularly benefit—they can review the written summary to ensure they understood verbal instructions correctly, reducing misunderstandings and improving case execution.
The searchable meeting archive transformed case continuity. When a case transfers between attorneys, the new attorney can search email for the client name and read complete summaries of every prior meeting discussion—understanding case history, client concerns, previous legal advice, and pending action items. What used to require hours of piecing together fragmented notes now takes minutes of reading comprehensive summaries.
The system also created legal liability protection. The firm now has objective, timestamped records of every client meeting discussion. If a client claims something was or wasn’t discussed, the transcript-based summary provides definitive documentation. This protects the firm from malpractice claims based on “he said / she said” disagreements about meeting content.

