At 9 p.m. Last Wednesday, I Was Still Writing Up Meeting Notes in a Café
I had just finished an online call with a client who does accounting and tax work. The hour-long recording was mostly back-and-forth confirmations. After getting home, I listened to another 20 minutes; my brain was fried, so I could only write down to-do items from memory. The next day, I missed a quote confirmation, and the client came back asking about it. At that moment, I thought: if something could listen first and pull out “who needs to do what, by when,” that would be enough.
Gemini Omni 1.2 Flash Is a Multimodal Model
Simply put, it does not just read text; it can understand audio, video, and images together. The official positioning is as a multimodal model for developers. For people like us who do not write code, it is more like hiring an assistant to listen to a recording and organize it for us. Who is already using it? The article does not list examples of ordinary entrepreneurs, and I have not yet run it through a full live client meeting, so I am not going to call it essential for everyone. I did use a similar workflow for one of my own review meetings: I sent the recording through an available Gemini interface, asked for a summary, decisions, and owners, then checked the output manually so a casual remark did not become a task.
If I am worried about raw material being uploaded, I would not immediately close the page. I might start with a public recording that contains no client information and check the current account’s privacy and data settings. This is not something every person needs, especially if you are already good at taking meeting notes.
Reproducing It Today: 0 Yuan, About 20 Minutes
Money: there is no single, consistent public price in the article, so I am not going to invent one. There may be a free allowance, but the page after login is the real source. Time: about 20 minutes to prepare the recording and check the summary. Technical barrier: no coding required, although the first setup for audio uploads and prompts may take some trial and error. First step: open Gemini’s official interface or the Gemini page I already use, find the audio upload option, and test it with one of my own public recordings. If I check the cost before paying, budgeting is easier to control.
If I Were Busy, This Is How I Would Choose
If I were just starting out, with only a few conversations each week, I would keep taking manual notes and try this once I started repeatedly missing items; there is no rush to test it now. With one or two clients, I would run a non-sensitive meeting clip through it and ask only for “items to confirm,” rather than sending anything automatically to a client. If I were scaling and the team met frequently, I would standardize the output into three columns: decision, owner, and deadline, while keeping human review in the loop. Whether to put it into a formal process would depend on using it for two straight weeks and seeing a stable error rate. This tool is not for everyone. Protecting client information comes before saving time.