Hi all, this is Isaac, I'm from Malaysia and I'm currently the Revenue Operations Manager for Oaky.com a hotel upselling software. As a non-native english speaker, Talktastic helped me turn thoughts into structured message, up my communication effectiveness across the team which is a big BIG thing consider Im the only team lead based away from the Amsterdam HQ and is incharge of the Ops component of the business. Hit me up in LinkedIn!
Isaac, you're the best. So glad we have you here. I'm curious about how you're using TalkTastic. Are you using it for emails, blog posts, short messages, or longer content? I'm particularly interested if you're using it at all for really short messaging or just for longer pieces.
Hey Matt, thanks for asking! I typically use TalkTastic for longer pieces of content. Sometimes these are responses where I just want to get all my thoughts out without worrying about filtering, structuring, or even thinking about the end result. For shorter replies or texts, I find it much faster to just type it out myself. This way, I can inject a bit of personality, be a little playful, and use some jargon or shorthand to keep it snappy. However, for longer communications like this reply to you, I prefer using TalkTastic to organize everything I'm saying. It can turn 30 seconds or a minute of my word jumble into structured content. I've been using this constantly for emails and other communication purposes because I think this is where TalkTastic really shows its value.
When do I use TalkTastic more? It's definitely for emails. It's amazing how TalkTastic recognizes the context on my screen and turns whatever I'm saying into a proper email format, so whenever I share this content, the email is always ready to be posted. - in my opinion this is the most accurate use case.
When it comes to blogging, technical writing, or jotting down instructions, TalkTastic might struggle to grasp the full background that extends beyond what's on my screen. For example, if I wanted to break down an entire project plan, concept, or multiple layers of context, it would be ideal to have a mix of text structure to work on those cases.
Interesting. What do you think we'd need to change or improve to make TalkTastic better - to make it faster than typing it out yourself for short messages? Have you tried using it for really short stuff? That's actually when it's the fastest, just so you know.
Hey Matt, I'm not sure I have a super comprehensive answer to this question. When it comes to shorter messages, I tend to keep things more casual and less formal - not so work-related, if that makes sense. Like, if I'm just updating a colleague about lunch plans or where we're meeting, using TalkTastic might actually make things sound too formal. Sometimes it's just quicker and more natural to type out those quick messages myself.
Imagine: when I was typing: WRU???? when I use Talkstatic: Where are you now?
Imagine: type: SFDC-Ordway load fail Talktastic: SF DC Artway, Luteville.
got it. Makes sense. I've been working hard on the context understanding to make it better at capturing the tone and linguistic style and to match it. See this thread: https://talktasticinsiders.slack.com/archives/C07QCS61WHM/p1730455101387269
For what it's worth, TalkTastic is actually a lot faster with short notes than long ones. And it's going to get even quicker. I'm putting a ton of effort into making the transcription short recordings super snappy and accurate. Oh, and just so you know, it automatically inserts the transcript for short notes; rewrite for longer ones.
For me, accuracy is more important than speed. Compared to my usual typing, TalkTastic is already a major upgrade, so I'm not too concerned about speed. Maintaining the current speed is good enough. However, what's really crucial is the accuracy, especially in context. I need it to ensure it's not getting overly elaborate, overly formal, or turning into a summary. That's what matters most to me. 🙂
In the screenshot above, I noticed a small keyword showing the time from output to send. I think one thing that's not clear enough in the current UI is understanding when the rewrite is complete. On my screen, I'm in dark mode and I have to look for a small blinking icon at the bottom left of the pop-up window to gauge if it's ready or not. Sometimes if talktastic, somehow over-edited my original context, I might jump between versions to pick out the best transcripts before copy-paste it, and send out. It actually takes more time between the output and sending. Nevertheless, 70 seconds and above worth of word rumbling got sent out in 7 seconds - that's just amazing! Though I think a lot of users, including myself, wonder why we're using 7 seconds instead of 2.
Slashed the above as the context was wrong from Talktastic
What I wanted to say is: I think a lot of users, including myself, we spend over 7 sec instead of just 2.67 sec to hit send from when the output is ready simply because the transcripted text is long and we are just reviewing the text before confidently sending it out 🙂
Well, we've got a bit of a complicated system right now for deciding which version to give you. We have an algorithm that tries to guess the most accurate version. That number you're seeing is the time to auto-paste - it's the auto-paste event from when you're done speaking. What we're counting is how long it waits until it injects the text into the text field. 🤓
As a RevOps, can I suggest different measures of measuring? From the perspective of correlation, perhaps what we want to measure is the time to send based on per second talk-time. (Edit: per 10 seconds maybe, per second might be too small to measure)
I like this idea.
if 20 seconds talk = 2.67 second to send = 1.33/10 sec talk time if 70 seconds talk = 7.41 second to send = 1.06/10 sec talk time in this comparison, which is a fairer metric, longer talk wins
Yeah, fair point. I've been wanting to develop a metric like this. Honestly, my goal is to get you responses back in under 900 milliseconds. That's the target I'm aiming for. This is great stuff. I need to figure out how to calculate that precisely in Mixpanel, but it's a fantastic idea. Thanks for this! 👍
You're welcome - you have a RevOps and a fans here. Please don't be a stranger to hit me up if you need ideas on data etc 😉
How good are you at Mixpanel?
Seen it but not used it. In my current setup, I extract product usage data via SQL then transform and analyse them on BI tools or spreadsheets. 😓