Every lawyer knows the feeling. You walk out of a 90-minute client conference, head straight into another matter, and by the time you sit down to record your time at 6 pm, half the details have evaporated. What exactly did you discuss about the settlement structure? How long was the section on costs disclosure? Did you cover the limitation period issue, or was that the other file?
Now multiply that by four meetings a day, five days a week. According to industry research, Australian lawyers lose an average of 10–30% of their billable time simply because they cannot recall what they worked on. That leakage translates directly to lost revenue — for a solicitor billing at $500 per hour, even a modest 15% leakage means roughly $150,000 per year walking out the door.
This is the problem that voice-to-time-entry technology was designed to solve. Rather than relying on memory or hastily scribbled notes, you can now record a meeting, upload the audio file, and receive professionally formatted billing entries within minutes. In this article, we will walk through exactly how that process works using LexUnits — from hitting "record" to exporting entries into your practice management system.
Why Meeting Recordings Are the Ultimate Billing Source
Before we get into the workflow, it is worth understanding why audio is such a powerful input for time billing. Unlike typed notes, which are filtered through your memory and inevitably incomplete, a recording captures everything: every topic discussed, every piece of advice given, every instruction received.
When you transcribe a meeting recording and feed it to an AI billing engine, the output reflects the actual substance of the conference — not your best guess hours later. The result is more detailed entries, more accurate time allocation, and fewer write-offs at billing review stage.
For firms that routinely handle matters in languages other than English — particularly Mandarin, Cantonese, or other Asian languages common in Australian commercial practice — audio-based billing solves another pain point entirely. Bilingual meetings often switch between languages mid-sentence. Manually reconstructing those conversations into English-language billing descriptions is tedious. An AI tool that handles multilingual transcription removes that friction altogether.
The Complete Workflow: From Recording to Export
Let us walk through the end-to-end process. We will use a realistic scenario: a 45-minute client conference on a commercial lease dispute, conducted partly in English and partly in Mandarin.
Record the Meeting
Use any recording device or app you already have — your phone's built-in voice recorder, a dedicated dictaphone, Zoom's recording function, or Microsoft Teams. LexUnits accepts all common audio formats including MP3, WAV, M4A, and WEBM. There is no proprietary hardware required.
For best results, place the recording device centrally so all participants are audible. Even with imperfect audio quality, the transcription engine handles background noise, accents, and overlapping speech reasonably well.
Upload to LexUnits
Open LexUnits in your browser and navigate to the audio upload section. Drag your audio file into the upload area or click to browse. You can add optional context before processing — for example, the matter number, client name, or any specific instructions such as "conference with client re commercial lease at 123 George Street, Sydney."
This session context is injected into the AI prompt and dramatically improves the quality of the output. The AI uses it to anchor billing descriptions with the correct matter details rather than generating generic entries.
Automatic Transcription
Once uploaded, LexUnits sends the audio to a transcription engine powered by Google Gemini. The engine processes the file and produces a text transcript. This step typically takes one to three minutes depending on the length of the recording.
Critically, the transcription supports multiple languages. If your meeting was conducted in Mandarin, English, or a mix of both, the engine transcribes each segment in its original language and the AI billing model then generates English-language billing descriptions regardless of the source language. You do not need to translate anything manually.
AI Generates Time Entries
This is where the real value lies. The transcript is fed to an AI model (powered by Anthropic's Claude) that has been specifically prompted to think like a legal billing professional. The AI identifies distinct billable activities within the conversation, estimates the time spent on each, and generates descriptions that follow Australian legal billing conventions.
For a 45-minute conference, you might receive six to ten individual entries. Each entry includes a professional description, a time estimate in six-minute units, and a confidence score indicating how certain the AI is about the entry.
Review, Edit, and Polish
The generated entries appear in an editable table. You can adjust descriptions, change time allocations, merge or split entries, and remove anything that should not be billed. LexUnits also provides a "Polish" feature — select any entry and the AI rewrites the description to sound more professional, without changing the substance.
The Thorough/Conservative toggle lets you control how many entries are displayed. In "Thorough" mode, you see everything the AI identified, including lower-confidence items. In "Conservative" mode, only high-confidence entries are shown, giving you a cleaner starting point for review.
Export to Your Practice Management System
Once you are satisfied with the entries, export them in the format your practice management software requires. LexUnits supports CSV export configured for Actionstep, LEAP, Clio, and Smokeball. The field mapping is pre-configured so you can import directly without manual reformatting.
Multilingual Support: A Practical Advantage for Australian Firms
Australia's legal market is uniquely multilingual. Firms in Sydney and Melbourne routinely serve clients who prefer to communicate in Mandarin, Cantonese, Vietnamese, Korean, Arabic, or Hindi. Immigration practices, international trade matters, and cross-border property transactions frequently involve meetings conducted entirely or partially in a language other than English.
Traditional billing workflows force lawyers to mentally translate these conversations before recording time entries. That extra cognitive step introduces delays and errors. With LexUnits, the audio goes in as-is, and the billing entries come out in English. The AI handles the linguistic bridge, so the lawyer can focus on the substance.
This is particularly valuable for junior solicitors who may be fluent in the client's language but less confident writing formal English billing descriptions. The AI produces entries in proper legal billing style, which the supervising partner can review without needing to understand the original language of the meeting.
What About Audio Quality?
A common concern is whether the transcription will be accurate enough if the recording quality is poor. In practice, modern transcription engines are remarkably resilient. They handle phone-quality audio, slight background noise, and speakers with strong accents.
That said, some basic practices help:
- Avoid recording in very noisy environments like busy cafés or open-plan offices during peak hours.
- If using a phone, place it on the table between participants rather than in your pocket.
- For Zoom or Teams meetings, use the platform's built-in recording feature, which captures audio directly from the digital feed and avoids ambient noise entirely.
Even when transcription is not 100% perfect, the AI billing model is trained to work with imperfect input. It focuses on identifying the substance of discussions rather than requiring word-perfect transcripts.
Time Savings in Practice
Let us put some numbers around the efficiency gain. A typical 45-minute meeting generates six to ten billing entries. Manually typing those entries — including opening your practice management software, recalling details, writing descriptions, and allocating time — takes most lawyers 10 to 15 minutes. If you have four meetings in a day, that is 40 to 60 minutes spent on data entry alone.
With LexUnits, the same process takes about three to five minutes: upload the audio, wait for processing, review the entries, and export. The net saving is roughly 30 to 45 minutes per day. Over a five-day week, that is two to four hours returned to either billable work or personal time.
But the more significant gain is in revenue recovery. Because the AI captures details you would have forgotten, the entries are more comprehensive. Firms that adopt AI-assisted billing consistently report a 10–20% increase in recorded time, not because they are billing more aggressively, but because they are billing more accurately.
Privacy and Confidentiality
Legal professionals rightly ask about data security when it comes to client meeting recordings. LexUnits processes audio files through secure API connections. Recordings are not stored permanently on LexUnits servers — they are processed, the transcript is generated, and the original file is cleaned up automatically. The AI models used (Google Gemini for transcription, Anthropic Claude for billing generation) do not retain or train on your data.
For firms with strict information governance policies, it is worth noting that the entire workflow happens over encrypted connections (HTTPS/TLS), and no client data is shared with third parties beyond the API calls necessary for transcription and AI processing.
Getting Started
If you bill by the hour — and particularly if you regularly conduct meetings that are difficult to bill from memory — audio-to-billing is probably the single highest-ROI feature you can add to your workflow. The setup takes less than five minutes: create a LexUnits account, upload your first recording, and see the entries generated before you decide whether it is worth continuing.
Most lawyers who try it once never go back to manual entry for meetings.
FAQ: Can I use any recording device or app?
Yes. LexUnits accepts MP3, WAV, M4A, and WEBM files from any source — your phone's voice recorder, a dictaphone, Zoom, Microsoft Teams, or any other recording tool. There is no special hardware or software required.
FAQ: How does LexUnits handle meetings in languages other than English?
The transcription engine supports multiple languages including Mandarin, Cantonese, and other commonly spoken languages in Australian legal practice. Regardless of the source language, the AI generates billing entries in English. You do not need to translate the audio or transcript yourself.
FAQ: Is my client's meeting data stored on LexUnits servers?
Audio files are processed through secure API connections and cleaned up automatically after transcription. LexUnits does not permanently store your recordings. All data transfers happen over encrypted HTTPS connections, and the AI models do not retain or train on your data.