Perplexity Computer for Counsel

0:00So, one of the first things, why don't we just get started with like um you know, easy mode is what I'm going to call it, like our web search example. So, let me go ahead and share my screen and give you a sense of like just as something as simple as like if you could imagine your job. Um, we're doing this for AI regulation because that's kind of how I think about it, but um it could be breaking news to the Supreme Court. It could be this, you know, latest um uh IP litigation that you're tracking um you

0:31know, whatever is moving in your world. And this can be global as well. It doesn't have to be limited to the US. It doesn't have to be limited to certain states. And I think all of us struggle with just the amount of information that's out there, where to get that information. And so I love this search just because it's going to just do it every morning at 9:00 a.m. So that's like it's scheduled for you. It's going to do what you're kind of curious about and then have it broken down the way that you want to see it. So, here's kind of like a little sample of us running

1:01this report. Um, this is our monthly AI regulation report. Gives me a sense of what's moving. Um, you know, I can kind of it has like key call outs like states that we should be paying attention to, which I love. But I saw a lot of people that are international. You could, you know, this is easily adaptable to your

1:19use cases as well. um maybe there's some key cases that are starting to pop in this area. You want to make sure that you're um at the water cooler or chatting with your partner or your colleagues and you're like, "Oh yeah, of course I know Starbucks v Google," which I don't, but now I do. Um and then like the nice thing about this is like it also kind of like resets your focus for the day of like, okay, given that's what's going on like externally, now I can use that to set up a meeting with somebody. Like I can use this to uh maybe do a deep dive like Florida versus

1:53AG. Never heard of that one before. So why don't you know like tell me more about that? Um so tell me more about that. And it would just basically like take it from there. Um so one of the things that's great people are always like well why can't I just Google that? It's like well perplexity is an AI powered search. It's using LLMs to basically combine with the power of web search. It's using natural language. So, I know all of us kind of are more accustomed to doing that web search using those keyword searches that we got that we grew accustomed to using that language that we used as the

2:28internet was coming up. And now you can just kind of use real real language and it's actually going to that natural language is actually going to like help you find the thing you're looking for a little faster instead of like trying to figure out like what the key word that's going to like unlock where you're trying to go. It can bring in so much more context for you. So I love that. Um so

2:49yeah, that's is our simple web search. Um so I'll pause there. Um Jenny, questions or you know should we go into the next one? Yeah, I just want to call out the sources tab. So there are between search and computer. A lot of folks who have used search before, there's always citations and computer's the same way. You can look through all the citations that it's sourcing um and click directly exactly where it's um taking the information from. So there's always a way to backtrack and and confirm, double check the information that computer is pulling from. Yeah, I think that's a great point because people are like, uh,

3:26how do I trust the AI? I don't trust the AI. like I have great judgment. Well, this doesn't substitute your judgment. This actually just allows you to do it in an organized way and then you can do what you already do using your lawyer brain to go and look at the sources and where everything came from and then you can kind of like actively decide like are you know how do I feel about these sources? How are they saying it? Can I

3:47you know take a critical lens to that? Um so that's absolutely what I um I love to take a look at this for and I point people to. Um, and um, I click through to a lot of these sources just because of the lawyer in me. So, great call out. And just in case, maybe let's just say

4:02that you don't trust aillegal.com. That's also just a follow-up that you can make with computer too to say please rerun this without these specific sources or please rerun this and double down on the specific sources that I I like. Yeah. You know, like I think Scotas blog is like a very wellrecoognized, well-known blog. It's like that's a for instance. I haven't tried it, but like you could say those are the types of places that I want to use and I think that that content there is very authoritative. So you can help it like kind of determine and it can be a back and forth dialogue with which sources

4:37it's pulling from. Nice. Well, that's web search. Should we do a fun one? Not that fun. Let's do it. All right. Uh, I was Jenny and I were taking um we're prepping for this and I think it can be really fun for non- lawyers for lawyers to think about like it's hard to think in your exact practice area sometimes like what

5:04the broader applications are of things. Um, and so we have this really great feature called model council which will basically allow you to kind of like if you're looking at your life it's on a chessboard it basically helps you like flip it. So, we we wanted to use like more of a silly example, like more of a pop culture example. Um, because this is like something that isn't really like knowable or a fact. And I think it helps like tease out like what's really great about this. Um, I think this also like gets to the point of like which model should I be using? Like I'm using Opus,

5:35but I don't know. Or like maybe I don't even know what Opus means. Like this is, you know, what's great about Perplexity is like you don't have to know what the best models are because we already do that work for you. you are using the best models like when you're on our platform, you only need that one subscription. And then you can draw and take a look at how different models treat information differently. Um which evidence that they might site. Then you can see like where they disagree like you're it's like oh you know all the AI it's all it'll just tell me this. It's like well they actually take different

6:05positions too which I think is really interesting. And I think it's also like it helps you kind of get comfortable and build trust with like the basis for some of the decision- making or the some of the like I shouldn't say decision-m but more of the like analytical process that they go through. Um because you obviously own the decision, you own the judgment, but it can kind of show its work a little bit more. The other thing I love about it is that yeah, like it can like uncover things like maybe one of them like uncovered something very

6:33interesting, have like a unique finding. Um but then it can like ultimately like draw on all this and give you a comprehensive like um answer which I think is really great too. Um my personal favorite and this is where my personal bias because um UCLA grad but I do think that LeBron should have stayed at LO in LA. So thank you GPT for that [laughter] uh or I guess it's Opus and uh Gemini 3.1. Um, yeah. I think the other thing I like to uh demonstrate with model counsel is when you have something that's like really challenging, just to bring it like more to a serious level and think

7:09about how it applies to legal work. When you have a problem, us lawyers, we like we're paid to basically see like the the the next move on the chessboard and the next six moves. And then you want to understand what your opponent's going to do as well. And this works in like contract litigation, negotiations, this, you know, it it really runs the gamut. Like we're all doing this all the time. And I think this also helps you develop like what are the counterarguments to my position because one of the things that we get I think just as humans is like you get really kind of like wedded emotionally to like

7:44one line of thinking and you start to like have that confirmation bias built in. So then now you can like basically flip the chessboard around. Maybe you were playing white, now you're going to play black. And then you're like, "Oh, oh, that's like where that is a weakness." Or I'm that allows me to anticipate what my like what someone I'm working with, negotiating with might be saying in response. Um, so I really love that. Like you can say, "Oh yeah, like I kind of knew you were going to go that way because D." And then that you can kind of come up with some fallback positions. Um, so you know, let's say

8:17you're a Supreme Court litigator. Hopefully there's some of those on the phone. But, you know, it's like you have your initial response, you have like what's going to happen as a follow-up, you have your challenge, like you can kind of go in um ready ready to go. So, yeah, that's our fun model council example. Jenny, anything else you wanted

8:34to call out? Yeah, I for the first one on a AI regulation updates, I don't know if you uh some folks have caught it that you can schedule this on a daily basis. So, we've seen some of our um customers, they set up a model council every day and they upload and connect their arguments or something that they're working on. And you can have council model council check every single day with my updated notes with my updated thinking and my memo that my team and I have been working on. keep poking holes in our argument and instead of people have historically opened up Claude, opened up GBT, opened up Gemini and type

9:13the same question in every single time. And this really just guarantees a 360 view um to Allison's point about flipping the chessboard and and seeing if you can catch any holes in your argument. Yeah, I think this is like really valuable. I find like at actually the end of the process too, like let's say you've been going deep on a research project and you're really trying to get to like the answer that makes the most sense for like your client like where the law is and then you're starting to go down like one road and then you're again the confirmation bias is like starting to like kick in a little

9:46stronger and you're like great, I've got it, I've got it, I've got it. Like this has to be the thing. And I don't know about you, but I know in my career, I've definitely like gone and presented like what I consider is like the most comprehensive answer I could have thought of, you know, like that ever was. I did all the work and then I brought it to like a client or a senior partner and they like immediately flipped it on me and caught like what I hadn't thought about. So I think this is like just another um great prep tool. Um and then I think it also does build

10:14confidence in how models are approaching information. So, if you agree with LeBron going to the Sixers or not, [laughter] let us know. But maybe you could try some others as well. Um, on computer, you can just click the plus sign on the omniar and you should be able to see model council to initiate all three. You can also choose the three different models as well. So um if you maybe specifically like Fable for deep

10:45thinking or opus um etc. Yeah, great point. Um I you know I think as you start to work in AI like more and more you realize that some of these models are better at certain questions. Um, and that you can also use it to like do some quality control or like you know, you know, again, once you get to the end, it's like you think you have the answer now go like double double check. Um, so yeah, I love that part about it. Um, cool. Should we move on to

11:16maybe another one? Yeah, maybe another skill. Okay. So, one of the things I love about perplexity is it can basically take a lot of my like I don't know us layers have lots and lots of information like kind of coming at you. You have you also have to think about what you already have. So you have a question and maybe you need to draw on what is like in your like drive folder or your SharePoint or like your maybe it's your company's um whatever repository you have and trying to search for that. I call that like dumpster diving for like digital like contracts sometimes or like you know

12:04trying to find a needle in a hay stack. You basically have all these different files. it's so cheap to like start a new document these days and so cheap to like set up a new file and easily can get things lost. So now you can kind of bring together multiple different sources and try to um kind of bring a comprehensive look without having to do as much of that manual labor that you would have had to do. So here we have an example of like you know you're trying to get something done. Maybe you're on

12:33your phone. Maybe you're moving quickly. Maybe you have meetings. You're in between a meeting and you're like, I know that there was something in my inbox. I know that we have a whole contract repository. I actually used this yesterday cuz I was trying to build a a new model template for something and I needed to flip a contract to somebody really quickly. And it's like, yeah, go ahead and like look at all of our prior red lines, our negotiation history. Um, make sure that you bring over some of like the key points that we like always push back on. Um, and then like you know take that summarize that all up and you

13:05know now we're going to have like some sort of re like uh like repetitive repeatable guidance. So yeah, like here we can kind of take a look at how we did that. Um, this is like a pretty pretty uh meaty prompt. Um, but yeah, this is I literally did this yesterday. Not this particular example, but I did do this and I it literally helped me come up with like a new perplexity template helped me move a lot faster. I didn't have to go ask our like our

13:38contracts person, our legal ops person. I didn't have to go through every line of my Gmail and try to figure out where was that email. It pulled both like the email, the context. It looked at all our previous red lines on similar contracts. Um, it actually took a little too far and I had to dial it back a little bit and I was like, "Okay, not like don't put out the most protective version of this. Maybe we dial back to like something that's like people are most likely to accept and like what we are most likely to live with." Um, and so now I have that and I can actually just

14:10like flip it out um right away and I can tell it to email that out. Um, and hopefully we get to um basically turning a signed like building a new template to having a signed contract in under 24 hours. Basically, one thing you mentioned that I want to call out is preferences. Um, there are preferences to be super strict. There are preferences to have specific

14:33language across the work that you do. And perplexity does pick that up and store it in its memory. Um, but there are ways that you could either turn it off if that is something um that you don't want to keep on. Maybe it's different client work or different types of projects. Um, but just wanted to shout that one out that it continues to

14:55memorize your preferences. Yeah, it gets better and better as you use it. And I think the reason that I find it very valuable is you don't have to bring so much context to every single individual prompt. It's going to start grabbing more context and then reusing that context like the more you call on it. And I think that's what Jenny is

15:15saying like it gets better and better. And I'm trying to like break down like how does it get better and better? Like it's like well it's just going to remember like I'm a lawyer. I'm at Perplexity. We've you know done this contract before. Maybe we've done that contract together like 50 times. Like it's going to start to remember like you know how you what you tend to call for and what you try to set as like the

15:36stage when you're when you're prompting. Someone did have a question on different models. So for this one um you can maybe see on the bottom right under the omniar it selected a specific model that's happening but what's happening in the back end is that it's distributing different models for the right task. So for research maybe the more recent model

16:01of um opus is better at doing research. Maybe a different model of GPT is doing better at summarizing and building it out into a memo. um it generally builds that out into the thinking steps. So a lot of folks are asking how do I know which is the best AI platform? We're constantly fine-tuning our backend to better understand what is going to be the best output. So if this is also by preference if there if you know and you've been testing out different models are better for some things you can get specific. But if not this is also where

16:37you can just trust perplexity. Yeah it's great. I love to experiment with different models. So what I will do um when you're starting like a new thread, you can actually go here and you can just say like pick me the best model. And what Perplexi because it has so much information about which what models are good at which tasks. It will route you to the right one for what you're asking about. And when I mean like which tasks, like some models are very good at coding and that they've been kind of trained to do that a little bit better. We don't us lawyers might not necessarily need that. We need the

17:12the models that are more attuned to like formatting um figuring out like formal context versus informal context. Um bringing all that like what is what's what's interesting and useful and required in the legal practice. Um those kinds of contexts and so it can kind of find what's the model. You don't have to play the should I go sign up for cloud, should I go sign up for anthropic or I'm like confused and now I'm stuck and you know you you can just start here as a good entry point and you will like naturally start to learn some preference. You'll start to have some preferences for models naturally over

17:47time even though you don't really um you don't have to know why. That's [laughter] true. I still don't know why. I'm like I don't know. I kind of like that one. Like maybe it's but if a new one comes out it takes Yeah. and a new one comes out and then you try that one out, you know. So, um

18:01that is pretty pretty fun. Um okay, I also saw someone ask what's the difference between search and computer? These are computer queries though the they um more of the research ones you can do you can deep dive deep with search. Um, but when it comes to actually building out something with the information, whether it be a dashboard or a app, um, that is where a computer is going to come in extra handy. And, um, I think we have a pretty cool one

18:33lined up next. Oh, great. Um, so this is like when I think about setting the stage, this is like the most beautiful thing to look at. So, I'm going to like let you guys look at this, but I'm going to explain like what are you looking at? Us lawyers have to sort through a lot of rules, a lot of rags. They could be state specific. Um, every state, county, jurisdiction, they may like house those rules differently. And they may be like in different parts of different code sections. And um you could see why this lent itself to like printing like literally everything out and then using

19:08a highlighter and tabbing everything and then trying to like coales it and bring it all together. What's great about this, this is just to give you an example and to get you thinking creatively about what how you might be able to apply this to your world. This is something that's useful for a litigator. I don't do a ton of litigation, but you know, I can immediately see like if I needed to know, let's say, privacy laws, like which ones were which states are moving privacy laws right now, like where would I be able to find the citations for that? Like which one has like the strictest rule? Um, you know, here we're

19:40doing it on based on like subpoenas and what's, you know, what are the different rules and you can see that there's a lot of variety. Um, you know, and then this can also, like just to zoom it out, this can work globally, too. I know a lot of us, especially if you're um working on working for um international companies or you're supporting international companies at an outside council law firm, we can get really stressed of like I don't know that area of the law or I don't know that country's particular legal regime. I'm being asked to appine on, you know, this jurisdiction. I didn't go to law school there. I don't

20:15practice that. Um, but I think what's great is that Perplexi can basically pull that all in, visualize it for you in a way that's like very digestible. And again, it's not going to replace your judgment. This isn't replacing anything. It's just giving you the citations and allowing you to see like what the rules actually are. And then to the extent it's making like, oh, this is like stricter, less strict, like you can go and interrogate that and kind of

20:39apply your own judgment on top of this. But I find this to be like a very novel and easy way to grab a bunch of different rules that may or may not conflict, bring them all together, visualize it, and then give you a little bit more of a road map for where you're going to go next instead of spending the entire time like, you know, this could have been like a a summer project that like an intern plus would have had to do plus a bunch of engineers to like make this visualization. Um, I know when I was at previous companies, I always had to like beg, borrow, and steal for

21:11engineering talent and like have them come and help me like make sense of like all these different repositories or rules or things that I'm trying to organize. Um, and Perplexity, I think, um, can do this. I think we did this in like a one shot. Um, [laughter] now you have to be super I mean this is not you know this is not the the result you're going to expect every time but like you know you're going to see you'll get better and better at being able to produce like this quality of thing um as

21:38you go. Yeah. Someone was asking where do you even start with computer to build this and it is all really starting from your curiosity and your um where you want to go with it and you're saying hey like what do here's who I am here's my goal here's who I'm trying to accomplish what I would really like is this information and um it would be best if I could oh I'm so sorry I got too still it my lights turned off um But it is there we go. Um but you can then kind of keep tweaking and following up and say hey I would actually really like this

22:19information um in this way. Perplexity will try to do its best to understand what you're trying to accomplish and make a maybe a few UI guesses. Um but yes we will be sharing the the query used to build out this dashboard. um as with all the queries in today's um workshop and hopefully that can give you

22:43some inspiration to build your own. Amazing. Yeah, this is I mean it's really impressive but you can also see like how generally applicable it is whether you know it doesn't matter what area of the law you're practicing in we all have to do this heavier work that isn't the most fun. Um, and so yeah, bringing in something that can just like synthesize a lot of data that's in a lot of different styles and in a lot of different places and then bring it together to basically build I'm going to go back to the chessboard analysis. Like this is going to like help you build your chessboard. Like what what rules,

23:20what are your constraints, what are you working with? And now you can start to see like where your strategy can come in. Um what is novel in your circumstance? like um it just kind of like sets you up to now apply your judgment, your experience um you know instead of I wasted all my time trying to like tab and highlight

23:41and like go find these things. Someone was also calling out the freshness of this dashboard and it is also just a matter of asking and following up hey can you update this every day. Um and if you connect it with some infrastructure tools like subbase or even publish it on berscell that can also make sure it's storing the information um if you also want it with it want it within a specific backup or um other security measures with that

24:10platform as well. It's another great point. So when I worked at a different company, I did do I had like a 50 state jurisdiction um I want to say I guess practice uh where we had to basically keep a massive spreadsheet. Um my brother's an accountant and so he's like, "Oh my god, never let lawyers like input into spreadsheets because all they do is just put in like so many words." But I mean that's like the that's like the best tool we have, right, to do a 50 state survey. But then those rules change all the time. So then who's who's

24:43responsible for keeping it up to date? And I think we've all had that experience if you've tried to organize 50 states rules that the first challenge is getting it all organized and the second is keeping it evergreen. So, I think it's a great point what Jenny is saying is like, yeah, you could build this so that it keeps it evergreen and you don't you're not like kicking out like stale giant Excel spreadsheets that looked at once and then they immediately became um stale and you can't use them anymore and then

25:14you know what a wasted effort. I think that to that point it sounds like a change log would be super helpful on a daily basis to know exactly where. So remember, perplexity is always using specific sources that you're either pointing it to or trying to see what is um what is the most referenceable and accurate. You can also ask computer in plain language to say please pull that out into a change log by state by date note when this was changed and let me know and let me see the article that it is sourcing it from. So that is just by

25:52um asking and following up. I think this gets us into like the other thing I wanted to show which is um basically like how could you do like a more like this is more of a snapshot right and you can keep it up to date but we also have this um for you which is like how we've thought about building something that is meant to stay perpetually up to date and is meant to help you um like find things like that are quick moving that require like real time tracking Um, I can't think of a better example than regulatory and policy work. You really do have to

26:30basically take that 50-st state map and then look at which state legislatures are moving, which laws have come into place, what regulators are saying. Um, and so, you know, you can basically build this for yourself without having to explain to engineers within your company or the IT department at your firm like this is what I need to be successful. this is what's going to make my life like so much easier. You don't you can basically be your own engineer and you can say I would like to have you know something going like this and keep things you know up to-date tracked. I want it all organized in a beautiful way

27:09that makes sense to me. I think there's been a huge gap in how we all use software in our personal lives. And then when I come back to like what legal teams have, I think there's been like a real drop off in terms of like the the creativity and engineering resources that have been applied to solving legal use cases. So again, I think that's what gets me like really passionate about perplexity. It's like yes, finally I I have the tools. I feel like, you know, superwoman to have like my brain plus perplexity going at uh these questions and basically building my own products

27:44for what I do on a day-to-day basis. Uh someone mentioned recently the the best idea the best suggestion to get started is build the worst version of your best idea. Oh, I think oneshotting is really overrated. Um it's pretty cool. explain what oneshotting is in case people don't know what that means. Yeah, oneshotting means you you submit one query or one prompt and then you get exactly what you wanted. That would be insane if all the context that you have in your files and your company and what they know about you and pulling all that context into a dashboard. What you don't see behind here is the iterating. Um

28:28maybe like after you start using a bit, you want to add a review cue. Um, after you want to pull an external intelligence, you could also follow it up. And it's really I keep thinking like what is your idea? And it's okay if it's a bad idea, but it needs to start somewhere. Um, and you can just keep fine-tuning it with the best thing is just in your plain language rather than how can I translate this to an engineer so they would under be able to

28:59understand. Yeah. I feel like breaking it down like starting from the big picture but then you can also ask perplexity to help you. You don't have to have the answer like how to do it because none of you know most of us didn't go and become product managers or product designers or we're not familiar with you know all the human factor work that goes into like making our apps really smooth and easy to use. That's

29:25okay because now you have perplexity. you can tap into that knowledge that's already out there that you just haven't been able to access or be aware of. And then you can say like, "Hey, I have this challenge. What I would like to see is and can you help me think through and build like what you like help me build and and improve this? What what what are questions you might want to ask me? What

29:48do information might you need from me?" and it will like start iterating with you and give you a sense and you know basic basically like brainstorm with you before it gets to like this level of like oh my gosh I have this like super complicated multi-layered app that's doing all the things um if this feels like too much and oftentimes it is very hard like Jenny said to do it all at once in one question. So start with, you know, your like more of an ideating space and think about more about like what's your biggest challenge? What what feels heavy for you today? What makes it harder for you to

30:24get your work done on a day-to-day basis? And frankly, I find like starting with a pen and paper is often like it's like what is my, you know, just to be able to write like what is my biggest challenge? You know, like it's like a little bit of a period of self self-reflection and then you can go back to perplexity to say all right, I'm ready to take this one on. Um, and let's go. and what you know you'll be so blown

30:45away by what you get back. Someone was asking about um connectors or files that you share with customers. We connect to all the different file apps. So whether that be Box, SharePoint, Dropbox um and like I mentioned earlier, it respects the permissions that you have that other users have of what files they have access to. So if that is a share shared file, you can also ask it to connect um there and same with data licenses as well. So I believe um maybe legal zoom isn't the best example for this crowd, but you can connect your legal zoom license. There might be different tools and you might have different access that

31:27will also respect your permissions as well. Yeah, there's like ways to connect um you know, Pacer or other uh Lexus Nexus like other subscriptions that you have. You can basically like put computer on top of some of your existing subscriptions, your existing um resources across your company that help bring information in. And it's really great at trying to like plug those in and talk to it. Um, and and take some of this manual work of having to like take a concept here, move it to another database, pull that you pull from that

32:02database and all bring it all together. Um, but Docu Sign and Box are like a great example. Drive, like even just being able to do that. I tried to tell the product team around here, I'm like, think simple lawyers. You'll blow lawyers away with just like being able to like link your email, your files, and oh my god, cherry on top, Docu Sign, you could push something out to somebody like and have it signed by them. um using your Docu Sign connector, using your Docu Sign credentials, um but h not having to leave your screen, not having to download something to upload something to spend the time

32:42putting all the little signature blocks in all the little right places. Um you know, I I think I had to sign a um a housing contract. I was selling a house and I needed to like have I don't know however many fields put down and and with initials and three parties and um I basically was like oh my gosh I I don't know this is like a whole day project I don't have time for and then you know threw it into perplexity and you know basically we got it done we got it done in a couple minutes. So um so yeah let me pause there. any other things on this

33:17or can we go into maybe using some of that proprietary data? We've got a good example. I think one thing um we we don't go into in this example, but it's not just reading the information that you have from connectors, but you can also write too. So, every different connector though for our direct integrations, we have settings on what you can push information to. So whether it be you have a updated contract, you could push that to docysine, get it loaded up. Um I update a lot of documents in my Google Drive and I ask computer once it does a review and provides comments to to push

33:54it to Google Drive. Um but speaking of documents, I think that would be great to show even if you do work with a lot of offline documents. I know some some places are a little bit more wier on connecting. Um or you have you're going back and forth offline on a specific agreement. Um you could also upload that

34:16as well. Yeah, it's a great point. We use this a lot. Um we often get like thirdparty paper for example and it comes over via email. It comes over in as a PDF. It comes over as a um a word doc. It's not in our files. we can't just call it. Um, and so we're downloading it, we're uploading it, and we can bring it to Perplexity and like immediately have it kind of take a pass at this. So, uh, yeah, we have a lot of tooling that's basically set up to do contract review and make contract review a little bit more efficient, um, rather than you

34:53having to kind of like, um, you know, be in a dark room in a basement with like no light to like really, really, really, really, really focus and make sure that, um, you know, where all the, you know, prior negotiation points were. Um, and then go over like with a fine tooth comb like making sure that there's no mistakes. Like this can kind of call this out for you. And then it's not doing it for you. You're not like skipping the part that you review the contract, but it's making it a little bit more fun. It's like you sent it to, you know, a junior associate to take a

35:23first pass and now it's going to flag something for you like some errors. Um, I love this because I I have a tendency to make a lot of typos. I feel like my analysis I spend so much time on the analysis and getting the legal argument right and then my typos are just glaring and it was like something I you know had to live through um all my years of practicing because as lawyers we actually have like a filter for thinking about things like if you have like one typo basically your document's trash like doesn't matter what the substance is and that puts a lot of pressure on

35:58us. um we feel a lot of pressure to get things right which is then what slows us down ultimately because you need to have a lot of brain space, you need to have basically like the time, the energy, you need to have like the right conditions to like really heavily focus on something. So, and and it could be like relatively simple, but your dayto-day could be kind of crazy that day and you're like, I know all of us, all of my friends, all of my co-workers are like, I I mean to do this thing, but I just need the head space. I just need to take 15 minutes

36:28when I have a second. Well, this is like the extra seconds you're getting back. Computer's taking a crack at it for you. And it's kind of like mapping out like where you might want to spend like the most amount of attention. Like that's like a glaring error. Like here you've, you know, this there's like some things that are like not, you know, like

36:44consistent. Like here we can flag this. This doesn't mean you have to accept all of these things. It doesn't mean they're all right. There could be things in there that you're like, but yeah, based on my experience, like that's not a problem. we can leave that in. Or like if I were to take that out, the other side would give me like so much more hassle than it's worth. Great. That's where you apply your judgment. We This is not a substitute for judgment. It's just meant to make it a little bit

37:08easier to apply that judgment. So here you can kind of like call your attention to things. I know, you know, perplexity often takes a first pass for me and they very much operate like a junior associate where it where it issues spots but doesn't solve the problems. And so a lot of it is like and

37:26so what like what if I did have this? Well those are the questions you can iterate with it too. So it doesn't mean like just because it did this one pass you can't go back and like keep working on it. Um but yeah I think we we find this to be like a really helpful tool internally particularly for people like me like I I always say I'm like good enough or dangerous enough in contracts but not like I am not a pro. I don't think, you know, and you can um go ask my previous employers like, would you want to pay this person to like do

37:55contracts all day? Um and I think the mutual answer to that is no. Um but I like know enough to be dangerous. Like I can do a contract here and there, but it gives me more confidence when I'm doing that that I'm not accidentally like overlooking something that um a more trained like person that does contracts like day in day out, has been doing it for 20 years would catch. Yeah, one thing here is that you'll see that there's only two words in the query and that's because we triggered a skill. Um, what it skill is is that you can condense your instructions into a skill

38:32and you can make it very specific to how you want it to be. So rather than brain dumping all of your context about who you are, what company you work for, what type of um severity you want this uh agreement to be looked at in um any way to personalize it, you would just be able to put in a skill. Um you could either do a forward slash or there's a little plus button that goes through your entire skill library and then it'll go through all the instructions that you had had it do in the beginning. So, it's not going to do it in a follow-up,

39:06unfortunately. But maybe if you want to go to the homepage, Allison, I go like, yes. So, at the bottom on the plus sign, you will be able to um Oh, I think it's because this is a demo account, so we don't know. Yeah, it's a demo account. I know that's the problem. I'm like, that's why I'm like, ah, I hate clicking around like um but basically what I would do is

39:29like, let's say I'm trying to do that. It's like let's build a skill which is just a fancy way to say let's build let's do something that's repeatable where you can like bring context for and then like think about optimizing for like a particular result. So that's what final pass essentially is is like our version of um just building you a skill that's like yeah build me I want to be able to have you do a final pass. I feel like I'm done with the contract. It looks pretty good to me now. I'm ready to push it out. what do you know do a final pass on it. Um and we had it

40:05basically go research what what makes for good contracts or what make and then you can say now research for what makes good contracts at my company. Um and then basically remember that and that's the skill. It just brings that memory that context to a new document a new example without you having to be like remember I am a lawyer remember that I have all these contracts. Remember you have to go like search them. remember this um it takes the even that

40:34friction out of the process. Speaking of repeatable things, I think we have um time to share just one last project. Oo, I love projects. Um I'm really excited to share this one. Um I've been going really big on projects lately and this uh we've had like uh what was it called before, Jenny? They were called spaces. spaces. So, you may be familiar with spaces if you're a longtime Perplexity user. If you're not, that's okay. This is basically like the V2

41:06version. Like this is the next level. Um, so yeah, we have like maybe you know you're going to be working on like one particular thing more intensely and it's going to have a lot of sub projects to it. So you can kind of like start to organize from the beginning and then when you're starting like new sessions, you know, like let's say you wanted to like do a new session here. Gosh, go back. Go back. You know, you can be like go like go look for context

41:38from my commercial litigation project. And now you have the power to ask something. bringing in all of this context here. And then you can also always edit the instructions. So your your mouse is hovering right over it. Maybe you can click the edit button to show Oh, it's my mouse. Sorry. Bottom right on the

42:02context side. Oh, yeah. Yeah. Um you can add context of instructions. You can add persistent files to this project. Also prioritizing specific links. Back to where I said you can focus on links. When you add links and files here, these this context will be prioritized in the search. Um there is a way that you can say I only want you to search these files and links or you can just have it prioritize as well as search the rest of the web and the

42:30files that you have connected. put it to like a real world example. We're using like commercial litigation and litigation is a really great example for this what what this does for you because you're constantly having to refer in litigation to like the previous ruling, the previous like order, the previous stipulation, like all those files that are generated just from one single case. You have to basically like work within the sandbox that is that case. And you also have to bring in like

43:00prior precedent from outside of that. But you really really need to think about like what happened in this case already. That's like some of the most important things. So then it can also help you find like the needle in the hay stack that's in some of those documents like gosh remember like maybe you're not

43:17using this even for a current project. Maybe this happened six months ago or two years ago. It's like what was that that case and we made this like great argument and it was the the most winning argument and then the judge loved it and then like cited it in her ruling. Like you can use that and that's what Kell what Jenny is trying to say is like you can have it like really focus on particular files and so you can call from those particular files to help answer some of the questions or to think about ways that you can like go crazy with you know this. Oh no, what have I

43:48done? [laughter] Oh no, there we go. Okay, this was one of the memos that it created. Um, and the great thing about projects, you'll see the little person icon is that you can share this with your teammates as well, so they can see what you're creating. Um, work on memos with you, see what the updated research is, so that everyone is working from the same type of

44:13information as well. Yeah, I love this. Um we are starting to use this a lot on our team. Um even just between me and like one or two other people who may be working on like the same question and we're trying to work at it on it like um and have and have context shared amongst the two of us or three of us whatever the team is so that we're like I already ran that search. I already grabbed all the contracts. I already did all that. I already grabbed you know and then that person can just from your work keep layering on. Um, and so it's a great I don't know

44:48collaborative environment for you. Like I think it's like going into that next level. I think a lot of people are used to having Google Drive or Box or someplace and I think that was a big game changer. Google Docs, Google Sheets, those all of those productivity like evolutions were great and they're really useful. But this takes it to a whole new level because it can do it across a bunch of those things. It can call all of those things at once. it can combine it with web search and now you can like basically use it to collaborate with your teammate who also wants to

45:21bring in this exact type of context. And speaking of layering, do you want to click through brain real quick in this project? Oh yeah. See if I can uh [laughter] Oh no. Maybe back and then click on that folder. Back back more. Yeah. And then click commercial. Okay. There we go. Yeah, there we go.

45:44And then you can click on brain. Brain. When you It's brain is like memory. Yeah, I love our naming convention. And what it'll do is that when you have multiple people working on the same type of project, it's going to develop and remember all these different concepts, these entities, and these work streams and how they all relate to one another. And this can also either call out um relationships with certain concepts that maybe you as an individual might not call out but as a teammate you have different with amongst teammates you can call and see the connection. So you can see this context graph on the lefth hand

46:23side you can even click into one of the entities on the dots above and then you can see a whole deep dive of how does it connect to each other? What does this mean? What does it own? And sometimes it doesn't always get it right or sometimes it doesn't have the full context. And that is where you can edit it on the upper right corner to say here's what they actually own. Um my midsize litigation boutique at my tiny size litigation boutique. This is um a fictitious law firm that I created. Um you can you can save that and it'll continue to remember this memory as

47:01well. And what I find really cool about Brain as well is that some some of our customers are firms and they have all these different um different customers and clients. Maybe you don't want that memory shared across your entire perplexity memory and you can keep this brain this memory contained in this specific project as well. So there's a a toggable toggable say that fast 10 times um toggable version to say I want it contained here or I want it to share um

47:35information across the board. Yeah, I love like basically it's a fancy way to say it it just remembers a lot more context about your particular project, your needs, who you are. And the more context that you can kind of preload or that you can borrow from previous assignments like it doesn't have to be um like I've started projects without having like without writing any of this or without writing any of that without bringing any files to it but I know it's something I want to call on as I go um and so you don't have there isn't like h I have to go build the brain or build this project. It's so

48:14instead you can start from wherever you're at and then as you're starting to find this helpful you can start layering in more and more things and you'll be surprised like ooh ooh now I have an idea now this is like feeling lighter right like this is an idea what if I dumped in all of these cases and then I wanted to look at this and you know I'm bringing this context um I think it can be really helpful and it works just as well if you don't have this all set up you can ask it basically to develop it for you you can ask it to um you can it

48:45iterate with it. It's not like, "Oh my god, I set up the brain wrong." Um, you know, and and it's very transparent how it's like bringing in different pieces of context. And so, again, as Jenny said, like if it got the wrong end of the stick about something, you can go and change it um and edit. But, um, it

49:03really does make your life a lot easier. Um, because then you can kind of start to like level up given like again, I'm going to go back to my chessboard metaphor, like chessboard metaphor because it's powerful. Well, you're basically like resetting the chessboard from the last time you were looking at that particular like in that particular game and you're going back to it exactly at the turn that you want to be at instead of like let me [clears throat]

49:28like reset the board from scratch. You're like it just immediately drops right into where you left off. You were about to checkmate Jenny and like then you got called away and now you're back, right? You're ready to like put the moves on her. um it can just bring all that in and then it helps you given it's basically like you get to kind of reflect back like what your challenges are and then once you kind of see them it helps you figure out what the next steps are and gives you new insights and like I I feel like a burst of creativity when I have a lot of context put into

50:00these like projects because then you're like ooh then you can start to help me think about like all of these constraints and all of these constraints and all these constraints and basically like your whole life right your whole life has a bunch of different pattern patterns and constraints that you're trying to bring in and the creativity is in that like negative space. And so that's I think what the like where projects come in. They can help you find that negative space where you are like oo I hadn't thought about that yet. Like that's amazing and you knew that I couldn't do this and you knew you

50:28couldn't do that. And just to like leave it on like a a really silly example, um I started using this on um in my personal life and I was like stuck on how are we going to get my three-year-old daughter out the door and um you know she hates getting her hair done and everybody's like so maybe do

50:47your do her hair the night before. Boom. Like I mean that's very very simple and really silly but honestly had never occurred to me before. And so this could be like the new way I've leveled up. That's the new negative space that I've found in my life. Um, new insight I've gained. And I think that that's like what's very powerful about this. It isn't just like a static file repository. It's putting you like right at the exact moment you're going to check me Jenny and you're like, what

51:19do I do now? What's the move? And remember, a lot of folks were like, how is this different? We are built on our web index that is built for AI. Um, and people are asking does it understand this system? Does it understand that specific M uh framework? You there is research out on the web of these specific frameworks and systems that you're already referring to. And it's always just saying, hey, go reference this. Maybe go look at this specific um this specific website that I know has laid out the principles the best. So that is where it's going to give you the accurate information and it's not going

51:57to just give you context from when the the model was last trained on. Um my my personal experience I actually got into a bike accident this past weekend and I immediately created a project and say please give me um legal and health advice best as you can sourcing the web um of how I how I can manage it. So it will also it'll automatically update that brain as you update it. I brought in my notes from the emergency room. I uploaded my X-rays health premium sources and that brain is

52:32going to update by itself. Um but I I hope you found this helpful both for work and for personal. Um I hope your daughter's hair is holding out throughout the night. I know we'll see if that's a fix. You know, there may be a new constraint soon. Toddlers are very good at finding constraints. But um but yes, someone did ask, we are we are HIPPA certified. I'm not just uploading my X-rays into the ether. Um and I do have I do have data retention configured as well for a few days. So um I we will be sharing the recording, we'll be sharing the queries, um we'll be sharing the

53:11commonly asked questions and a list of FAQs. Um, so I I hope you enjoyed it. Please follow up and look out for more sessions. But thanks so much for being a great crowd, asking amazing questions. Um, Allison, thanks for being a great presenter and thank you for organizing everything. This is always so fun, especially when the lights turn off at

53:34the [laughter] end. All right, I'll leave it to you there. Um, but have a great week, everyone. Thanks. Bye.