Latest / Inside DuckDuckGo / Duck Tales: How we're making it easier to use Duck.ai, our private AI chat. (Ep.40)
Transcript
- 0:01SPEAKER_00: Hello and welcome to DuckTales,
- 0:05SPEAKER_00: where we go behind the scenes at DuckDuckGo and discuss the stories,
- 0:09SPEAKER_00: technology, and people that help build privacy tools for everyone.
- 0:14SPEAKER_00: In each episode,
- 0:15SPEAKER_00: you'll hear from employees about our vision, product updates, engineering approach,
- 0:21SPEAKER_00: And in this particular episode,
- 0:24SPEAKER_00: you're going to hear from Mihai,
- 0:27SPEAKER_00: who is here to talk about the various UX improvements that we've made
- 0:33SPEAKER_00: to DuckDuck.ai recently.
- 0:37SPEAKER_00: Mihai is on the design team.
- 0:39SPEAKER_00: And do you want to just give a quick introduction of yourself, Mihai?
- 0:46SPEAKER_01: Sure.
- 0:46SPEAKER_01: Thanks for having me.
- 0:49SPEAKER_01: My name is Mihaly.
- 0:50SPEAKER_01: I'm originally from Moldova,
- 0:52SPEAKER_01: lived across the world in many places,
- 0:55SPEAKER_01: now based in Switzerland, joined DuckDuckGo maybe around three years ago.
- 1:00SPEAKER_01: And I've been on the DuckAI team almost from the very beginning.
- 1:05SPEAKER_01: So I've seen the product evolve since then dramatically.
- 1:10SPEAKER_01: So yeah, we'll talk about it today.
- 1:12SPEAKER_00: Thanks.
- 1:13SPEAKER_00: Well, thanks for joining.
- 1:14SPEAKER_00: For anyone who hasn't met me, I'm Bia.
- 1:17SPEAKER_00: I'm on the product team here.
- 1:20SPEAKER_00: I've been around for a while.
- 1:22SPEAKER_00: So yeah,
- 1:23SPEAKER_00: I'm going to go ahead and ask Mihai some questions about his work on DuckAI,
- 1:27SPEAKER_00: and we'll just dive on in.
- 1:30SPEAKER_00: So I think I said we're talking about improving DuckAI UX,
- 1:35SPEAKER_00: and I'm not sure I said that if you're not familiar,
- 1:38SPEAKER_00: UX is just a fancy abbreviation for user experience,
- 1:46SPEAKER_00: even though we all know experience doesn't start with an X.
- 1:53SPEAKER_00: This means making the features that we have and the functionality
- 1:57SPEAKER_00: that we have more usable,
- 1:58SPEAKER_00: tweaking the functionality to better suit people's needs and desires,
- 2:06SPEAKER_00: and just make the product more delightful and easier to use.
- 2:12SPEAKER_00: Feel free to amend that definition.
- 2:14SPEAKER_00: But just to set the tone a little bit, Um,
- 2:18SPEAKER_00: Mihai, tell me a little bit about, like, how we even decide to work on something.
- 2:23SPEAKER_00: Like, how do we prioritize, uh, UX user experience changes in DuckAI land?
- 2:31SPEAKER_01: Nice.
- 2:32SPEAKER_01: Um, yeah,
- 2:33SPEAKER_01: as designers,
- 2:34SPEAKER_01: we're basically, I,
- 2:36SPEAKER_01: I tend to think of myself as the diplomats that we,
- 2:40SPEAKER_01: we try to talk to as many people to get our, um, work done.
- 2:45SPEAKER_01: Uh, but we use various signals to inform these decisions.
- 2:50SPEAKER_01: Um, some of, some of them are more perfect.
- 2:52SPEAKER_01: Some of them are imperfect.
- 2:54SPEAKER_01: Um, there's no There's no reality where we're working with perfect information.
- 2:59SPEAKER_01: So sometimes we form assumptions.
- 3:02SPEAKER_01: Sometimes we rely on research, user feedback, internal feedback.
- 3:07SPEAKER_01: Sometimes we have strategic bets that we place As an organization,
- 3:13SPEAKER_01: we have dashboards with metrics.
- 3:19SPEAKER_01: We know the competitive gaps that we have.
- 3:22SPEAKER_01: And sometimes we just rely on instinct.
- 3:24SPEAKER_01: So we bundle all of these together in our decision making process.
- 3:29SPEAKER_01: Some of these signals They surface a problem,
- 3:34SPEAKER_01: others validate the direction,
- 3:35SPEAKER_01: and it's on us to use our own human judgment to determine the right tradeoff
- 3:41SPEAKER_01: for these things.
- 3:44SPEAKER_01: And in the end,
- 3:45SPEAKER_01: it narrows down to looking at where the biggest overlap is with the user problems
- 3:50SPEAKER_01: that we have and where we're taking the product.
- 3:53SPEAKER_01: Yeah, maybe this is more kind of like general response, but Yeah.
- 3:59SPEAKER_01: No one decision was the same.
- 4:00SPEAKER_01: So it's all based on experienced people in the room making the calls based on
- 4:07SPEAKER_01: the information that we have.
- 4:09SPEAKER_00: Yeah.
- 4:09SPEAKER_00: I mean, one thing that I'm guessing people,
- 4:12SPEAKER_00: I bet a lot of people would underestimate the degree to which we're obsessed
- 4:15SPEAKER_00: with user feedback.
- 4:17SPEAKER_00: Like, you can't completely rely on user feedback to tell you how to build a product
- 4:20SPEAKER_00: because,
- 4:21SPEAKER_00: you know,
- 4:23SPEAKER_00: Hopefully the people building the product have additional insights
- 4:26SPEAKER_00: about what's possible and what good looks like that users can't articulate
- 4:29SPEAKER_00: or shouldn't take the time to articulate if not on them.
- 4:32SPEAKER_00: But that aside, we obsess pretty hard on feedback.
- 4:36SPEAKER_00: We read an insane amount of feedback.
- 4:40SPEAKER_00: We have systems for categorizing it and making sense of it.
- 4:43SPEAKER_00: I've been caught up in one discussions where we...
- 4:47SPEAKER_00: Are really deeply trying to interpret a comment somebody made on Reddit or in
- 4:51SPEAKER_00: the App Store and like try to figure out the best way to address it.
- 4:54SPEAKER_00: And so I'm just pointing that out because I think that might not be obvious.
- 4:58SPEAKER_00: Like I think there's probably a lot of,
- 5:00SPEAKER_00: you know,
- 5:00SPEAKER_00: for a tech company that serves millions and millions of users,
- 5:04SPEAKER_00: I'm not sure that's, I'm guessing that's not the norm.
- 5:08SPEAKER_00: So that is a big part of it.
- 5:09SPEAKER_00: But then like beyond that,
- 5:11SPEAKER_00: you know,
- 5:11SPEAKER_00: there's a bunch of other signals we use,
- 5:13SPEAKER_00: like Mihai said,
- 5:14SPEAKER_00: and I know in DuckAI,
- 5:16SPEAKER_00: on that team,
- 5:17SPEAKER_00: like you guys have a rubric, right,
- 5:19SPEAKER_00: that you're using to actually like evaluate in a structured way,
- 5:23SPEAKER_00: a bunch of Project ideas and feedback is one of the inputs to that rubric,
- 5:30SPEAKER_00: but there are others as well.
- 5:32SPEAKER_00: So also very maybe distinctive of DuckDuckGo is,
- 5:36SPEAKER_00: I think it's a hyper-structured and rational decision-making process.
- 5:40SPEAKER_00: Would you agree with that, me and I?
- 5:43SPEAKER_01: Oh yeah.
- 5:44SPEAKER_01: Um, it's, it's quite unique.
- 5:45SPEAKER_01: Um, I think the,
- 5:47SPEAKER_01: the great thing is that a bunch of us,
- 5:49SPEAKER_01: everyone here, um, they're scoping their own projects.
- 5:53SPEAKER_01: They have their own ideas.
- 5:54SPEAKER_01: They uncover different things that they're pitching in and that goes into that rubric.
- 5:59SPEAKER_01: Um, and at certain times we,
- 6:02SPEAKER_01: those categories of how, um, How we scored the rubric, they're weighted.
- 6:07SPEAKER_01: So sometimes certain elements are more important than the others.
- 6:12SPEAKER_01: And that help us kind of look at what's important to us right now
- 6:17SPEAKER_01: and what we're getting from the users and all the projects
- 6:20SPEAKER_01: that are strong candidates for us to tackle in the next roadmap, for example.
- 6:26SPEAKER_00: Cool.
- 6:27SPEAKER_00: Yeah.
- 6:27SPEAKER_00: So let's talk about something that did bubble to the top recently.
- 6:30SPEAKER_00: Like, can you give me an example of a UX improvement that you've worked on?
- 6:36SPEAKER_01: Yes, so the first example is dictation.
- 6:40SPEAKER_01: I'll share my screen real quick.
- 6:45SPEAKER_01: Cool.
- 6:45SPEAKER_01: So this is our interface here.
- 6:48SPEAKER_01: So users keep asking for dictation features.
- 6:52SPEAKER_01: So it's this microphone in the input field.
- 6:56SPEAKER_01: There are certain people that for them is, they would use the input field rather talk than type.
- 7:05SPEAKER_01: It's either a prompt is really long or they just want to do it hands-free.
- 7:11SPEAKER_01: And shipping this was actually great because it got us in line with the competitors as well.
- 7:18SPEAKER_01: But internally,
- 7:18SPEAKER_01: we were kind of, well, it was,
- 7:23SPEAKER_01: the group was split in terms of like,
- 7:24SPEAKER_01: if we want to do it, until that user feedback kept coming back, kept coming back.
- 7:30SPEAKER_01: And we decided to pull the trigger.
- 7:34SPEAKER_01: From UX standpoint, what does it What's exciting for me personally is voice interfaces.
- 7:40SPEAKER_01: They have a big affordance problem that it's really hard sometimes to know what's
- 7:46SPEAKER_01: the system status for that.
- 7:48SPEAKER_01: And to avoid the jargon is basically,
- 7:50SPEAKER_01: you need to know what the app is doing at any point in time.
- 7:56SPEAKER_01: And that was an interesting challenge on top of just delivering your feature.
- 8:02SPEAKER_01: And if I trigger it, we do three things here.
- 8:06SPEAKER_01: Um, we label things clearly for people to know what's going on.
- 8:11SPEAKER_01: You also get a visual cue, this waveform.
- 8:15SPEAKER_01: That provides feedback.
- 8:17SPEAKER_01: As you see me pause, that waveform reacts to my voice.
- 8:21SPEAKER_01: But also the input field has changed its shape,
- 8:28SPEAKER_01: not the height or anything,
- 8:29SPEAKER_01: but what's inside it now is geared towards a different input, which is not text.
- 8:36SPEAKER_01: And we've added this nice glow animation to it, which is really cool.
- 8:44SPEAKER_01: Then it's transcribing and giving back the text.
- 8:52SPEAKER_01: Another one, should I give a few more?
- 8:57SPEAKER_00: Give me another example, please.
- 8:59SPEAKER_00: That's good.
- 8:59SPEAKER_01: All right.
- 9:01SPEAKER_01: Another one is batch deletion, which is another feature that was requested by users.
- 9:09SPEAKER_01: This one is a little different because we're famous for our fire button.
- 9:14SPEAKER_01: You can basically burn all your data and stuff like that.
- 9:22SPEAKER_01: But in DuckAI specifically,
- 9:24SPEAKER_01: we had The reality that you could either delete one chat at a time or all of them
- 9:33SPEAKER_01: and nothing in between.
- 9:35SPEAKER_01: And we decided that we need to give people more precise control to do it in just one step.
- 9:43SPEAKER_01: Whenever you have 30 chats,
- 9:45SPEAKER_01: for example,
- 9:46SPEAKER_01: and you just want to delete seven at a time, it's quite annoying to go one by one.
- 9:53SPEAKER_01: So this one It's not groundbreaking or anything.
- 9:58SPEAKER_01: It's been a pattern on the web for quite some time, how you can batch select different things.
- 10:05SPEAKER_01: This is a great example, thinking back to the signals that we spoke.
- 10:08SPEAKER_01: It's a great example when you Need to know where to follow the process
- 10:14SPEAKER_01: for something whenever you're designing something that's more unknown.
- 10:18SPEAKER_01: And whenever you need to trust your instincts more.
- 10:20SPEAKER_01: And this is a project that was basically that, where we used our experience more.
- 10:28SPEAKER_01: And the project was short.
- 10:29SPEAKER_01: This was delivered and shipped quite quickly because of that.
- 10:35SPEAKER_01: And what the feature does, you can go in and select
- 10:40SPEAKER_01: How many chats you want to delete?
- 10:42SPEAKER_01: We have different entry points to how to do it.
- 10:45SPEAKER_01: Usually we know people click the fire button to open it.
- 10:48SPEAKER_01: So we surface a button there for people to actually not delete everything,
- 10:55SPEAKER_01: but select multiple ones.
- 10:58SPEAKER_01: Or for maybe power users,
- 10:59SPEAKER_01: whenever something's selected,
- 11:01SPEAKER_01: you can shift click and enter this mode and delete your chats.
- 11:07SPEAKER_01: So yeah,
- 11:08SPEAKER_01: this one,
- 11:09SPEAKER_01: the reason I'm asking is we spoke about all these signals on how complex things are,
- 11:16SPEAKER_01: but this one was three designers.
- 11:18SPEAKER_01: We got in the room, we did a 30 minute thing.
- 11:22SPEAKER_01: Then I spent maybe two hours on it on edge cases.
- 11:26SPEAKER_01: We put it to review to a wider Group of stakeholders,
- 11:31SPEAKER_01: and it went straight into implementation,
- 11:33SPEAKER_01: which is great because we have features that are a little bit more complex,
- 11:37SPEAKER_01: which require all of that kind of digging research and all those other things.
- 11:42SPEAKER_01: But this is a great outlier, maybe, where we can just roll with it differently a little bit.
- 11:50SPEAKER_00: Yeah, yeah.
- 11:51SPEAKER_00: Also, there's Yeah,
- 11:54SPEAKER_00: while you're talking,
- 11:55SPEAKER_00: I was thinking about being that we also added not too long ago,
- 11:57SPEAKER_00: another another way to delete chat levels sort of at your cadence,
- 12:03SPEAKER_00: which is like when you hover.
- 12:04SPEAKER_00: Do you want to show your screen again real quick?
- 12:09SPEAKER_00: When you hover over.
- 12:12SPEAKER_00: Each chat in your chat history, you get a fire button.
- 12:15SPEAKER_00: So you can also just tap that.
- 12:17SPEAKER_00: So if you want to do one at a time, but from a list, that's another way to do it.
- 12:21SPEAKER_00: I think that was a nice win.
- 12:23SPEAKER_00: And then,
- 12:23SPEAKER_00: I don't know,
- 12:24SPEAKER_00: just for a little real time inside baseball,
- 12:27SPEAKER_00: truly an hour ago,
- 12:29SPEAKER_00: Mihai and I were on a thread where we were talking about with some other folks,
- 12:36SPEAKER_00: what the best equivalent of that on mobile is, where you don't have hover.
- 12:41SPEAKER_00: We just have to, if we want, I personally love and use that per chat fire button.
- 12:45SPEAKER_00: You're like, oh, I don't need that, I don't need that.
- 12:47SPEAKER_00: I'm going to clean it up real quick.
- 12:50SPEAKER_00: On mobile, we don't have that.
- 12:51SPEAKER_00: So yeah, like, is it worth the real estate to just always show the fire button?
- 12:55SPEAKER_00: Is that too noisy or people don't misclick it?
- 12:58SPEAKER_00: So we were kind of debating that this morning and going to get it into
- 13:01SPEAKER_00: an internal build and like kind of feel it out ourselves.
- 13:04SPEAKER_00: And then, you know, if it feels good, we'll release it, but we'll watch it.
- 13:09SPEAKER_00: Any feedback coming in as well to help guide us on whether it's working as we hoped it would.
- 13:17SPEAKER_00: Hot off the presses.
- 13:19SPEAKER_00: All right.
- 13:20SPEAKER_00: Do you want to talk about one more?
- 13:23SPEAKER_00: We talked about maybe the model picker might be interesting to show because that was
- 13:26SPEAKER_00: a nice and nuanced recent UX improvement.
- 13:32SPEAKER_01: Yes, let me share my screen again.
- 13:35SPEAKER_01: Yeah, this one is not the one I designed,
- 13:38SPEAKER_01: so I don't want to upset the designer who actually done the hard work.
- 13:43SPEAKER_01: But I advised this project, so I stayed close to it.
- 13:47SPEAKER_00: You want to name drop, give him a shout out?
- 13:49SPEAKER_01: It's Esteban, so thank you, Esteban, for a better model picker.
- 13:53SPEAKER_01: We're all grateful for that.
- 13:57SPEAKER_01: Yes, it's basically our team has grown, so it's great that we can cover more ground.
- 14:02SPEAKER_01: And Esteban is not new on the team, but whenever I started, it was just me.
- 14:08SPEAKER_01: So it's great to see all of these things basically ship
- 14:14SPEAKER_01: and all these amazing features that We can put out much faster.
- 14:19SPEAKER_01: Now, this one is basically,
- 14:21SPEAKER_01: speaking of old stuff, the old mobile picker had a whole lot more details in it.
- 14:29SPEAKER_01: And through research, we've uncovered that people just could not understand what's in there.
- 14:34SPEAKER_01: And they were stalling, trying to make sense of it a little bit.
- 14:38SPEAKER_01: Power users have no problem.
- 14:42SPEAKER_01: Kind of like general public people who are actually not experts on just want to use the tool.
- 14:50SPEAKER_01: It was a thing that was giving them pause.
- 14:53SPEAKER_01: And another thing that we've uncovered,
- 14:56SPEAKER_01: which was useful insight is once they got past that confusion,
- 15:01SPEAKER_01: they really valued it.
- 15:02SPEAKER_01: And it was the thing that they would recommend to friends and stuff.
- 15:07SPEAKER_01: And on top of it,
- 15:09SPEAKER_01: we also,
- 15:09SPEAKER_01: as a company,
- 15:11SPEAKER_01: DuckAI specifically, we provide all these LLMs from different providers.
- 15:17SPEAKER_01: So this is some sort of our own differentiator where you get the private versions
- 15:23SPEAKER_01: of all of these.
- 15:25SPEAKER_01: So we wanted to fix this old problem of people stalling and being
- 15:30SPEAKER_01: a little bit confused about it.
- 15:32SPEAKER_01: And the solution was basically Again,
- 15:37SPEAKER_01: based on insights,
- 15:39SPEAKER_01: we realized that people knew what they wanted to use or had some general idea
- 15:45SPEAKER_01: and they just use what's giving them a good response and they just stick to it
- 15:50SPEAKER_01: and majority don't really switch.
- 15:51SPEAKER_01: There are use cases for some that switch for different particular reasons,
- 15:56SPEAKER_01: but for this particular model or now a dropdown,
- 16:02SPEAKER_01: they just needed something much simpler.
- 16:05SPEAKER_01: And the job of this is To show that there is choice because we provide all
- 16:10SPEAKER_01: of these things, variety, but also we wanted to remove the guesswork for people.
- 16:16SPEAKER_01: So Esteban has done an amazing job of condensing this into showing that variety.
- 16:22SPEAKER_01: But we switched what we are doing here for the user is we're providing now
- 16:27SPEAKER_01: a recommendation of what we think and what we think that They'll use to have
- 16:34SPEAKER_01: a great time and a great experience on the platform.
- 16:38SPEAKER_01: For each subscription tier, we have free Plossom Pro.
- 16:44SPEAKER_01: We recommend two models.
- 16:47SPEAKER_01: One always good for everyday use.
- 16:50SPEAKER_01: Based on their tier,
- 16:51SPEAKER_01: and another one which will give even maybe a stronger,
- 16:55SPEAKER_01: more comprehensive response,
- 16:57SPEAKER_01: but we'll use more limits or it'll exhaust those limits faster, basically.
- 17:05SPEAKER_01: So yeah,
- 17:05SPEAKER_01: now it's kind of,
- 17:06SPEAKER_01: it's much simpler and much more enjoyable,
- 17:11SPEAKER_01: at least for me and based on your research for users to use.
- 17:17SPEAKER_00: Yeah, one detail is like before it was a button that you press and it opened a modal like this.
- 17:22SPEAKER_00: It opened a new screen effectively with your choices and then you select your choice
- 17:29SPEAKER_00: and then click Start New Track.
- 17:31SPEAKER_00: And like even just,
- 17:32SPEAKER_00: it's a small thing,
- 17:33SPEAKER_00: but like the difference between what feels like kind of a dropdown and something
- 17:37SPEAKER_00: that feels like a new screen where you have to pick and then say, start new chat.
- 17:40SPEAKER_00: Like I knew how it worked and it's not like one click takes so much time,
- 17:45SPEAKER_00: but the sort of like cognitive friction of like picking and
- 17:50SPEAKER_00: then clicking something again, like start new chat.
- 17:52SPEAKER_00: And like usually I hadn't even started a chat.
- 17:54SPEAKER_00: Like it just, it didn't feel easy peasy to switch around.
- 17:57SPEAKER_00: You know,
- 17:57SPEAKER_00: it felt like higher friction than you needed to So everything else was adding
- 18:02SPEAKER_00: and just kind of moving that,
- 18:03SPEAKER_00: like changing the UX pattern, I think it's a nice little win.
- 18:11SPEAKER_00: Sweet.
- 18:12SPEAKER_00: But that's one that probably is like counter to your batch delete example.
- 18:15SPEAKER_00: Like we really did kind of agonize over a bunch of nuances because like
- 18:19SPEAKER_00: with any design decision,
- 18:20SPEAKER_00: there's trade-offs and there's different cases that you have to account for
- 18:24SPEAKER_00: in different states.
- 18:25SPEAKER_00: And like it took a while to reason through it all.
- 18:29SPEAKER_00: Speaking of which,
- 18:30SPEAKER_00: can you give me, Mihaly,
- 18:32SPEAKER_00: an example of like a difficult or not obvious design decision that you've had
- 18:37SPEAKER_00: to make recently?
- 18:41SPEAKER_01: A recent one?
- 18:43SPEAKER_01: We always,
- 18:43SPEAKER_01: I think,
- 18:45SPEAKER_01: as a designer here,
- 18:48SPEAKER_01: I'm always reminded of how little we know about our user in comparison
- 18:53SPEAKER_01: to everything else we've experienced.
- 18:56SPEAKER_01: I remember like,
- 18:57SPEAKER_01: we just don't know the count of how many users we have just
- 18:59SPEAKER_01: because we don't track them and stuff like that.
- 19:01SPEAKER_01: Everything is very much on those guesstimates and kind of like,
- 19:06SPEAKER_01: people sometimes use phrases like,
- 19:08SPEAKER_01: I'm confident in this assumption, like 80% and stuff like that.
- 19:12SPEAKER_01: They give a percentage and it's a wonderful constraint because it's why I joined.
- 19:20SPEAKER_00: Because it's privacy respecting is what you're getting at.
- 19:26SPEAKER_01: Exactly.
- 19:27SPEAKER_01: And that goes with everything we touch.
- 19:30SPEAKER_01: Whenever we're designing it,
- 19:33SPEAKER_01: you design something and then it's through that lens of respecting the data
- 19:41SPEAKER_01: and providing a great experience without needing all of those other things,
- 19:46SPEAKER_01: which other companies do.
- 19:50SPEAKER_01: I maybe don't have a very good example to give that is recent.
- 19:56SPEAKER_01: One that is maybe an older one is whenever you're getting a response,
- 20:00SPEAKER_01: how you're switching through different models,
- 20:02SPEAKER_01: which is something which is currently being worked on and where you will be evolving
- 20:07SPEAKER_01: that feature soon.
- 20:09SPEAKER_01: But that one What wasn't too obvious about it is that switching models is not
- 20:15SPEAKER_01: a new paradigm in the user's mind because this AI thing is somewhat new.
- 20:21SPEAKER_01: That's why it takes the shape of a chat interface because it's familiar
- 20:25SPEAKER_01: and switching a model
- 20:27SPEAKER_01: In other places we've seen very much has different pagination approaches
- 20:33SPEAKER_01: and stuff like that,
- 20:34SPEAKER_01: which is now learned, but not natural based on what we've uncovered.
- 20:40SPEAKER_01: And every time when I spoke to people,
- 20:42SPEAKER_01: like I've seen so many confused faces about like the pagination of answers
- 20:48SPEAKER_01: and stuff like that.
- 20:49SPEAKER_01: And what What we did,
- 20:52SPEAKER_01: which wasn't obvious,
- 20:53SPEAKER_01: we basically took a more linear approach
- 20:57SPEAKER_01: to how those responses are getting surfaced,
- 20:59SPEAKER_01: which feels more natural into a chat interface,
- 21:03SPEAKER_01: a thing that we've been using for decades.
- 21:05SPEAKER_01: So we kind of went against the grain there.
- 21:09SPEAKER_01: And it wasn't obvious at first.
- 21:10SPEAKER_00: So what would have had a native response look like?
- 21:12SPEAKER_00: If we went that route, we would have opened a whole new chat or something?
- 21:18SPEAKER_01: No, it creates the same response,
- 21:20SPEAKER_01: but it gives you like a counter, like a page for the responses.
- 21:24SPEAKER_01: And then you paginate and it gets forked in different streams,
- 21:28SPEAKER_01: almost like a decision tree that you can paginate back and continue conversation
- 21:31SPEAKER_01: in different.
- 21:32SPEAKER_01: So that's even like...
- 21:35SPEAKER_01: I spoke to engineers who use AI.
- 21:38SPEAKER_01: They use Courser.
- 21:39SPEAKER_01: They use all these advanced tools.
- 21:42SPEAKER_01: And for some, it wasn't apparent that this feature worked that certain way.
- 21:48SPEAKER_01: So even advanced people who are used to these tools were a little confused by
- 21:54SPEAKER_01: the conventional norms, so to speak.
- 21:56SPEAKER_01: So it made me feel really proud a little bit because we kind of took a different approach.
- 22:03SPEAKER_01: I think we get more than 300k uses on the week from that feature, which is cool to see.
- 22:11SPEAKER_00: Nice.
- 22:11SPEAKER_00: All right.
- 22:13SPEAKER_00: Last question, Lehi, before we sign off.
- 22:15SPEAKER_00: Can you preview any upcoming improvements that people should look forward to?
- 22:24SPEAKER_01: It's the project I just wrapped, which is still being worked on.
- 22:29SPEAKER_01: So the feature may still evolve,
- 22:31SPEAKER_01: but one thing we're trying to add to the platform is, are you seeing my Figma yet?
- 22:39SPEAKER_00: Yeah, I can see.
- 22:41SPEAKER_01: Okay, perfect.
- 22:42SPEAKER_01: So this is very, very rough.
- 22:46SPEAKER_01: Figma is...
- 22:47SPEAKER_01: Hiding some of it.
- 22:49SPEAKER_01: Basically,
- 22:49SPEAKER_01: we want to add the share functionality to your responses,
- 22:54SPEAKER_01: something that we don't do currently.
- 22:58SPEAKER_01: And we want to do this at the response level, but also at the entire conversational level.
- 23:03SPEAKER_01: And once you click share, you're basically sharing a response.
- 23:07SPEAKER_01: You create a public link that will leave on our encrypted server
- 23:12SPEAKER_01: with encryption keys and all those other things that engineers are working on.
- 23:17SPEAKER_01: To make that super secure and everything.
- 23:19SPEAKER_01: And then you can copy and send it to people to get that response.
- 23:28SPEAKER_01: And you'll be able to see as someone who opens the link, you'll be able to see what was shared.
- 23:34SPEAKER_01: And you'll be able to delineate what's yours if you choose to reply or if you choose
- 23:40SPEAKER_01: to add it to your own chats.
- 23:42SPEAKER_01: So this is how it was designed.
- 23:45SPEAKER_01: Sometimes we evolve the work as we play with it live when it's in code.
- 23:51SPEAKER_01: But it's something that should help a lot of people send a link as opposed to send
- 23:56SPEAKER_01: a bunch of text.
- 23:58SPEAKER_00: Nice.
- 23:59SPEAKER_00: That eggshell white is looking pretty sharp.
- 24:03SPEAKER_00: Is that what it's called, eggshell?
- 24:06SPEAKER_01: It's eggshell 10, specific.
- 24:09SPEAKER_00: Eggshell 10.
- 24:10SPEAKER_00: Nice.
- 24:11SPEAKER_00: Awesome.
- 24:12SPEAKER_00: All right.
- 24:12SPEAKER_00: Thank you, Mihaly.
- 24:13SPEAKER_00: Anything you want to add before we close this out?
- 24:17SPEAKER_01: Oh, this is great.
- 24:19SPEAKER_01: Yeah, this is great.
- 24:20SPEAKER_01: Thank you for having me.
- 24:21SPEAKER_00: Yeah, thanks for being on DuckTales.
- 24:25SPEAKER_00: All right.
- 24:26SPEAKER_00: Thanks, Mihai.
- 24:27SPEAKER_00: Thanks, everyone.