Capture Unbilled Hours with AI-Powered Time Tracking

Most small law firms use the built-in time tracking in their practice management software like Clio or PracticePanther. These tools provide manual stopwatches and forms for entering time. While better than a spreadsheet, they still require attorneys to remember to start and stop timers or reconstruct their day from memory, which is where unbilled hours originate.

Consider an associate at a 5-person litigation firm. They spend 25 minutes reviewing an opposing counsel's email, checking two case law citations in Westlaw, and drafting a 3-paragraph reply. They get interrupted by a partner's call and the Clio timer was never started. At 5 PM, they try to reconstruct the day and bill for 'reviewing and responding to opposing counsel correspondence,' logging 0.3 hours (18 minutes) and losing 7 minutes. This happens multiple times a day, across five attorneys, accumulating into significant revenue leakage.

Some add-on tools try to solve this by prompting users. For example, Time Miner can scan calendars and call logs. However, it operates on metadata, not content. The tool sees you had a call with 'John Smith' but has no context on the matter. It cannot analyze the content of a PDF brief to suggest a time entry for 'legal research on motion to dismiss for case #1234.' The attorney still does all the cognitive work of connecting the activity to the matter.

The structural problem is that these systems are databases with timers attached, not context-aware agents. Their architecture is designed for data entry, not data inference. They lack the ability to process unstructured data like the body of an email or the text of a legal document. To do that, a system needs a connection to a Large Language Model API, like Claude, and a pipeline to OCR, classify, and summarize documents, components not native to standard practice management software.

The engagement would begin with an audit of your firm's data flow. Syntora would map where work happens: your email server (e.g., Microsoft 365), document storage (e.g., Dropbox, AWS S3), calendar, and practice management system (e.g., Clio). This 3-day audit clarifies API access and identifies the most valuable data sources for capturing billable time. You receive a scope document detailing the proposed connections and a fixed build timeline.

The technical approach would be a FastAPI service that listens for events. New emails would trigger an AWS Lambda function to parse the content using the Claude API. The API would extract the sender, recipient, matter ID, and summarize the activity. A similar flow would process new documents stored in AWS S3, using an OCR library if needed. This process takes under 500ms per document. The service uses Supabase as a PostgreSQL database to store draft time entries, ensuring all data remains on infrastructure you control.

The result is a daily digest of draft time entries delivered to each attorney for approval. Instead of a blank timesheet, they see pre-filled entries like 'Reviewed 12-page PDF brief from opposing counsel on Smith v. Jones' with a suggested duration. Attorneys approve or edit with one click. This system would reduce daily time entry from 30 minutes of recall to less than 5 minutes of review. The full Python source code and a runbook are delivered, giving you complete ownership.