AI models don't retrieve facts like a search engine. They use the input data (transcripts from your recordings), to write clinical notes.
The right posture isn't blind trust. It's informed vigilance. Here's what to do when you spot an error.
Step 1: Check the Audio Score and Review the Transcript
Start with the audio quality score in VetNotes. If it's below 0.96, audio quality is likely the root cause. Poor audio leads to an inaccurate transcript, which leads to inaccurate notes. Open the transcript and read through it. Does it reflect what was actually said?
Step 2: Press the Thumbs Down Button
At the bottom right of any note, press the thumbs down button. This downvotes the note in the algorithm and helps VetNotes improve over time. It takes two seconds and genuinely makes a difference.
Step 3: Identify the Error and Search the Transcript
Pin down exactly what's wrong (a drug name, a diagnosis, a treatment detail), then keyword search for it in the transcript.
There are two likely outcomes:
You find the term in the transcript. This is a transcription error. VetNotes misheard the audio and the note followed from what it thought it heard. The fix is improving audio quality: move the mic closer, reduce background noise, speak clearly.
You don't find it in the transcript. This is an inference error. VetNotes inferred something from the broader context that wasn't actually said. Inference is a useful feature, but it's a double-edged sword. The fix is to be more explicit in your spoken word. If a medication isn't being prescribed, say so. If a diagnosis is being ruled out, make that clear out loud.
The Most Common Cause: Audio Quality
The single most common source of AI scribe errors is poor audio. If you're seeing mistakes regularly, start there. A small improvement in recording conditions can make a large difference in note accuracy.