Hey everyone, I’ve been trying out Otter.ai for sales call notes and I’m still pretty new to it. I was excited to see the new 'action items' feature, but I’m running into a weird issue.
It keeps creating tasks that weren’t actually discussed in the call. For example, in a recent client chat about a pricing question, it generated an action item for me to “send contract details by Friday,” but we never agreed to that. Has anyone else seen this? I’m worried it might add confusion to my workflow if I can’t trust the auto-generated tasks. Any tips on how to make it more accurate? 😅
Oh that's a bit concerning. I haven't used Otter's action items much myself yet, but if it's inventing commitments you didn't make, that could really mess with client trust. I wonder if it's maybe picking up on hypothetical statements? Like if someone said, "Well, if we had the contract details by Friday, that would be great," and it misinterpreted that as a firm task.
Have you tried looking at the transcript right next to where it generated that phantom task? Maybe there's a pattern in the language that triggers it. I'm curious if turning down the "aggressiveness" of the feature would help, if that's even an option.
That's a specific and frustrating problem to have right out of the gate. It undermines the core benefit of automating that kind of note-taking. I haven't used Otter's new feature myself, but I've seen similar issues in other platforms that try to infer intent from conversation.
You might try reviewing the transcript and flagging that specific line where it misidentified the task. Some systems learn from user corrections over time. It's also worth checking if there's a confidence threshold setting for the feature, allowing you to make it less prone to guessing.
—HR