Is it possible to tweak the AI for Meeting Summaries?

[Michael_Leugs]

As a long-time devotee of note-taking, I'm loving the AI Meeting Summaries feature, but the AI could be doing a much better job at writing. If it were a truly chat-based AI, I'd be able to simply tell it what to change, and hopefully it would learn from my preferences for future session – but I'm not seeing that capability. (Can Zoom confirm this?) Are there any hidden settings to tweak the AI's note-taking skills? A few pet peeves:

• Names are shown as last-name followed by a non-grammatical comma. Better guideline would be first+last name on first mention, first name only on subsequent mentions IF only one user has that first name – and lose the superfluous comma.
• Most calls include some pre-agenda social chatter that really doesn't need to be summarized for the record. Is the AI smart enough to understand what constitutes social chatter and stop summarizing it?
• Action Items are usually pretty good.
• The AI could learn from Strunk & White on crafting concise sentences. While generally grammatically correct, there are plenty of opportunities to cut words and overall length – boosting readability.
• The AI uses a lot of passive voice and fancy verb tense constructions when a simpler tense would do. We do attendees a favor when we summarize their colloquial language use in the form of clear, declarative sentences.
• I suspect the AI seeks to fill a certain target length and number of subheads, even on calls that don't contain a lot of substantive content. I've got a lot of meetings to support, so conciseness counts. (See prior reference to Strunk & White.) 

Is there a do-it-yourself way to address this kind of thing in settings, or in dialog with the AI?

[stueker]

Dear @Michael_Leugs ,

yesterday, at Zoomtopia, Zoom announced AI Companion 2.0 that also features an add-on that allows to customize the AI Companion.
https://news.zoom.us/zoom-introduces-ai-companion-2-0/

[stueker]

Dear @Michael_Leugs ,
did that reply answer your question? If so, could you please mark the reply as a correct answer?