Well done on the execution as ever! One thing that would be helpful for me—generally when watching your progress videos—is understanding how the problem was framed and by whom. Did a group of users express that the current speed wasn't fast enough for them? I ask because I'm a bit unsure about introducing the idea of an accuracy-to-speed trade-off. Is the difference in speed significant enough that someone would choose to sacrifice accuracy for speed?

Wow, you folks are on fire! Just to confirm, am I understanding this correctly? Here: Based on my mental model, thinking of privacy controls: i have either allowed Storytell to summarize all content for me, or I've decided to keep it turned off until I choose to turn it on. Are you suggesting that this notification pops up when I have chosen to turn it off, as a prompt for me to turn it back on? If so, this redefines what "off" means because truly "off" means no prompting. I understand that permanently off is tricky because people may turn it off forever and forget about it. If I'm understanding correctly, maybe a language rewrite would make more sense for me as a user: the message should state that "Storytell is completely off, do you want to turn it on now?" (nod to privacy controls https://www.storytell.ai/support/share-your-feedback/feedback-please-privacy-controls-within-the-chrome-extension) This brings up another point: when does this notification pop up? Because as a user, if I'm okay with ST summarizing everything, then it's happening automatically. And if it's off, it's because I don't want work to be done. Is there a content length threshold past which the notification pops up, so it isn't always popping? As a user, if I don't have ST open but it's allowed to summarize and my summary is complete, do we need to rely on a notification overlay that may potentially get in the way? Could it be more discrete and contained, like https://capture.dropbox.com/qmnOoPuNXcYKZ17u ? n = 1... curious to know what other users think. Regarding the 2x video watching, I personally lean on YouTube's innate feature for that. Regarding Jing's comment, I didn't know that the footnote link pointing directly to a video timecode was possible because I thought the summary might draw from different parts of the video. But if that isn't the case, then yes, I love that feature! Great progress!

Super interesting, and I like that you're breaking away from the right margin. Curious to see how much traction this feature gets since I didn't originate the request. While I can see myself primarily using the "summarize" feature within this bubble, I appreciate how you're casting a wide net by making those four functions available. Can we track usage of the four functions specifically within the highlighted text context, separately from what goes on within the right margin? Also, I'm unsure how the bubble pops - is it by simply highlighting the text? By the way, on the topic of "highlighting", I tend to highlight the summary output more than the original text, and often copy and paste AI outputs into Notion or Evernote. Great work! ps: <3 your puppy

Ryan: as always, your solutions are very clean and elegant! As a user—and this may just be another way of expressing the desire for control—I would prefer a level 0 binary toggle (not hidden behind hamburger) because I am constantly visiting new sites and don't feel inclined to get into the business of managing blocked site lists. I would be curious to A/B split this binary approach against your more robust vision and learn what others think! Lastly, I am someone who uninstalls extensions myself when I don't want them to run 🫢😳 when it comes to PII corporate misbehavior has eroded trust, even under trustworthy conditions! So maybe a little message assuring me, when the button is off, that ST isn't looking at the data, and a design that makes it look like it's completely turned off and inactive, would help!

I just installed version 1.1.0 and I loved the efficient implementation of the functions for previewing, editing, and using prompts in a limited space. The prompt library feels particularly valuable to me as a user; I think, in general, folks are starting to recognize prompting as a true skill. I didn't mind losing the "history" tab, curious to know what others think of that. Great job! By the way, look at the ad on my screenshot :)

yes, and, theoretically, the system should detect/proactively offer a template based on the content—hypothesis being, most content is indicative of a meeting type

fabulous progress! thought starters, when we talk about output templates, which taxonomy are we envisioning? https://capture.dropbox.com/MjlTTHkYrdg0fnPW 1'05" at this point the output template sounds like column A on the screenshot above. 1'46" at this point it sounded like column B on the screenshot above, i may be projecting. i seem to lean towards this option, although that is a testable proposition. 2'26 either way, when chat is open, the prompt library felt really useful to me. part of the UX challenge pertains to not necessarily knowing what type of output i'll get from a prompt without experiment first—i.e., if the library suggests 'fibonacci scale', or 'gradient of agreements', i may not know what i'll get until i try. and this comment may be of even more relevance as the platform grows into outputting formatted prompts, like these (arbitrary examples) https://capture.dropbox.com/3vlZocyLzt8O2pZ7 or https://capture.dropbox.com/hkdRXicE2iMiOyAk or https://capture.dropbox.com/YMLmEQhBbO9wnahu at which point having the marketplace organized by output as opposed to prompt may make more sense. (i recently bought an image prompt on https://promptbase.com by perusing outputs. i saw the output i liked, and unlocked a prompt. i would not be able to reverse-engineer the prompt by looking at an output. i know text is a different game but the logic may still apply to both formatted and unformatted output templates) what are your thoughts? subscribed to this and want to see how it develops!