Devshi Mehrotra on AI, justice, and public defense

America’s criminal justice system relies on public defenders to represent anyone without access to a private attorney. But public defenders are chronically under-resourced and overburdened with caseloads. Can technology help lessen that load and, in effect, make the criminal justice system more just? This week’s guest, Devshi Mehrotra, says yes. Devshi is co-founder and CEO of JusticeText, a startup developing AI software to help lawyers search and review case material more efficiently and enhance police accountability. She joins Reid and Aria to talk about JusticeText’s offerings and opportunities, along with public-private partnerships and local governments’ evolving attitudes toward AI and innovation. Plus, a public defender shares her experience using JusticeText and Devshi demos the platform. For more info on the podcast and transcripts of all the episodes, visit https://www.possible.fm/podcast/ Topics: 3:10 - Hellos and intros 6:47 - What inspired the creation of JusticeText 10:38 - What America’s public defenders are up against 12:35 - How JusticeText can help 14:32 - On being a for-profit company 18:11 - Creating a new market 20:47 - JusticeText’s customer base 23:33 - Midroll 23:41 - A testimony from a public defender using the platform 24:40 - Requests for new features 25:58 - JusticeText demo 28:50 - How AI is supporting, not replacing, public defenders’ jobs 30:23 - The intersection of government and technology 32:53 - How local governments are thinking about AI 34:59 - Other technologies that could improve the justice system 37:28 - What entrepreneurial advice Devshi would give her younger self 38:43 - The future of JusticeText 40:42 - Rapid-fire questions

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:01Speaker 00

  A lot of public defenders love being able to show up as, you know, friends and confidants to the clients that they're serving. And unfortunately, just the grueling nature of their work takes away from a lot of that fulfillment many times. And so the way we think about it is how can technology help them double down and focus on the work that drew them to this profession in the first place?

• 0:34Reid Hoffman?

  Hi, I'm Reid Hoffman.

• 0:36Aria Finger?

  And I'm Aria Finkler.

• 0:38Reid Hoffman?

  We want to know what happens if, in the future, everything breaks humanity's way? What we can possibly get right if we leverage technology like AI and our collective effort effectively?

• 0:50Aria Finger?

  We're speaking with technologists, ambitious builders, and deep thinkers across many fields. AI, geopolitics, media, healthcare, education, and more.

• 1:00Reid Hoffman?

  These conversations showcase another kind of guest, whether it's Inflection's Pi, OpenAI's GPT-4, or other AI tools. Each episode, we use AI to enhance and advance our discussion.

• 1:13Aria Finger?

  In each episode, we seek out the brightest version of the future and learn what it'll take to get there.

• 1:19Reid Hoffman?

  This is possible.

• 1:28Reid Hoffman?

  Our criminal justice system relies on public defenders to advocate for those who cannot afford their own private attorney. But as you may well know, the system is overburdened and, at certain junctures, systematically flawed.

• 1:45Aria Finger?

  Public defenders are chronically given more cases than they can handle. As an example, The Atlantic reported that some public defenders in New Orleans had 19,000 cases per year at one point. That translates to seven minutes per client. This means that sometimes innocent people are encouraged to take plea deals and spend time in jail. Harvard Public Health suggests somewhere between 55,000 and 110,000 people could be wrongfully behind bars.

• 2:14Reid Hoffman?

  So how can we give our public defenders more time and resources? How can technology lessen the heavy burden that public defenders are carrying? And how does this ladder up to making a more just system for more people?

• 2:29Aria Finger?

  Our guest today is working day in and day out towards these answers. She is the co-founder and CEO of Justice Text, an AI search tool that condenses hours of body cam footage, video calls and jail calls into a transcript so that public defenders can highlight key moments and understand a case more quickly. Justice Text gives public defenders back the time that they so desperately need. As listeners may know, I'm incredibly passionate about this space. I can't wait for you all to learn about this much-needed service.

• 2:59Reid Hoffman?

  Here's our conversation with Devshi Marautra.

• 3:10Aria Finger?

  Devshi, I'm so excited for this. I want to kick off by reading a part of your bio. I feel like I know you pretty well. We followed you along in this journey. And then I read it, and it says, "She has accumulated extensive experience applying deep learning algorithms to protect the progression of retinal diseases, train robotic agents, and perform natural language processing on legal corpora." So we'll dig into the NLP, of course, and legal corpora later in the episode, but tell us a little bit about retinal diseases and robotic agents.

• 3:42Speaker 00

  My first introduction to the world of AI was in 2016. This was the summer after my freshman year in college. I had accepted an internship opportunity at a research lab out in Beijing, China. I was working with this incredible lab that was focused on like more fundamental biology and physics research. And the lab director at the time, he was really excited to have me on board because he had heard of this exciting new thing called deep learning. And he said, you know what, she's a computer science wizard. She can help us out with processing and analyzing some of these cancer cell images that we're collecting a little bit more efficiently.

• 4:24Speaker 00

  And so he kind of gave me a very high level overview of what he wanted me to accomplish. And little did he know at the time, I had truly only taken an introductory Python and C course, and I was a little bit overwhelmed. But this was my first summer. This was my first internship. I was committed to doing whatever I could. And so I kind of just dived right in. I started researching the very fundamentals of machine learning, making my way over to gradient descent, to learning how neural networks work, how you come up with training and validation data sets to convolutional neural nets. And honestly, like, the more I learned about this field, the more I was just blown away.

• 5:09Speaker 00

  It was crazy to me that you could just string together all these mathematical functions and accomplish all these really, really incredible and complicated tasks. And so by the end of that summer, I had a working prototype for them that would take in all the cancer cell images that they were collecting and produce, you know, the image segmentation. And so that was exciting. But I think the bigger takeaway for me for that summer was, holy crap, like, I've like discovered this whole new field that I don't think a lot of people are talking about at the time. And I want to pursue this further. I want to see where I can take this interest. And so the following summer, I interned at Google Brain,

• 5:53Speaker 00

  which seemed like the Mecca of deep learning at the time. I was so excited to have that opportunity. It was just cool being able to hear from the research scientists directly that I had been reading about for so long. After that, I went to Microsoft Research. I went to DeepMind. And so I was, you know, just had this incredible opportunity to work on some projects that were more NLP focused, others were more computer vision focused. There were also, you know, amazing work that was being done in robotics. So that's kind of what that line in the bio is referring to. But honestly, I'm just like, that was like the foundation and the genesis for me eventually doing the work that I am today.

• 6:35Reid Hoffman?

  Yeah, and it was the right tour of some of the most interesting labs doing this. So it doesn't surprise me that that created a good springboard. So now on to AI and legal, not just because I'm an investor. I think Justice Text is a great example of using AI for good. So I'd love our listeners to get acquainted with its origin story. What inspired you and your co-founder to create Justice Text? And say a little bit about what it is as well.

• 7:10Speaker 00

  So Leslie and I, we met on our very first day as undergraduate students at the University of Chicago. Both of us were studying computer science. We were in the same house. We also did the same pre-orientation program. And so there were a lot of different touchpoints for us in the very early days that eventually led us to becoming co-founders. But it's just really cool to reflect on that because the pre-orientation program that we did together was actually focused on social justice and community leadership. So topics that are pretty different from what we actually studied for all of our, you know, computer science coursework. But I think it set the tone for how I spent my time as a college student in so many ways.

• 7:55Speaker 00

  Three months into our freshman year on campus is when the Chicago Police Department released the dashcam footage documenting the police murder of a 17-year-old unarmed black American boy. His name was Laquan McDonald. And if you look at that video, it's truly horrifying. Laquan was the same age as both Leslie and myself. And I think in that moment, I just reflected on the fact that as a 17-year-old, I was moving to a new city, you know, starting my college experience, meeting all these cool new friends and

• 8:37Speaker 00

  Laquan, for no reason that anyone could justifiably come up with, had been taken from this world for being born poor and black in Chicago. And I think for me, one of the things that I spent a lot of time reflecting on and just learning about in the next couple of years was this history of racialized policing in the city. And it was really cool to see how my local community started to organize against a lot of this injustice.

• 9:12Speaker 00

  I would, you know, go to the local church and just see how people were coming together, organizing after-school programs. I would see how, you know, there were so many teachers and pastors who would just come together and fight for justice in the wake of a local government that had failed to provide that safety and security for them for so long. And I think a lot of those conversations and just the people that I was meeting was really inspiring. Going into my senior year, I knew that I had to do something. I had no idea what that would look like. What I had studied, the skills that I had, were as a computer science student. And so what Leslie and I ended up doing was cold calling our local public defender's office,

• 9:54Speaker 00

  basically telling them, hey, guys, we're two college students with extra time on our hands. We'd love to help you guys out however we can. And honestly, in that very first conversation, they told us that we, as public defenders, are getting absolutely flooded with the volume of body-worn camera footage, interrogation videos, jail calls that we're receiving in our day-to-day caseloads. And we don't have a good system for getting through that. You guys are two computer scientists, maybe you can help us out. And so that was the genesis of Justice Tech. This was really just a side project that we built for a handful of attorneys in our city. And over time, realized that the problem is so much bigger than what we initially realized.

• 10:39Aria Finger?

  Well, I love it's just such a typical startup story where you're meeting customers and you're serving an unmet need. And you know, you guys were activists and not instead of, but in addition to your fury about what was happening, you said, let's figure out sort of a more systemic solution. So for those of us who aren't familiar with sort of the day to day work of public defenders, can you go into a little bit more, maybe like a case example of what sort of they're up against? And then when Justice Text comes in, how is that helpful?

• 11:07Speaker 00

  Public defenders are representing low income individuals who otherwise cannot afford an attorney. In many ways, they are that last line of defense to help this individual navigate this grueling bureaucratic criminal legal system that we have in America. And unfortunately, public defenders today are grappling with incredibly high caseloads. To the point where in some jurisdictions, they estimate the average public defender only has around 10 minutes to spend on an individual case. And due to that, this issue that we stumbled upon with body cams is really fascinating because in addition to these

• 11:51Speaker 00

  high caseloads, all of a sudden, over the past 10 years, these attorneys are also receiving hours and hours of body worn camera footage of these jail calls, of 911 calls as part of each case. This footage can be incredibly powerful. Previously, an attorney would have to just rely on what was written in the police report and kind of take that at face value. All of a sudden, they have access to this data that can help them corroborate that information and actually get to the truth of what may have transpired during that police interaction. But it's very hard to say that all this footage is actually leading to greater accountability if they don't have the tools, resources, time, infrastructure to actually

• 12:34Speaker 00

  be able to go through it. So that's why we built a solution that essentially allows attorneys to upload all of that data onto Justice Text. Within minutes, they'll get an automated transcription that they can search through. They'll get summaries of all that information. We will look out for key moments. When was someone brandised? When was someone administered a field sobriety test? When were they placed under arrest? Did the client ever request a lawyer? And then attorneys can use that platform to create video exhibits that they actually use in court to potentially impeach an opposing witness or to use as part of their opening or closing presentation.

• 13:12Speaker 00

  And so if I had to share an example, there's an attorney out in Northern California who had received a case where her client's vehicle had been searched. Her client was Spanish speaking. And, you know, in the police report, it had just been written as, you know, we searched the car, we found XYZ, that's why this individual is now detained. When she actually looked at the video, when she uploaded it to Justice Text, what she was able to identify was her Spanish speaking client had never been read his rights. And in fact, the officer had indicated to the English speaking friend to just tell him to go along with it. And that was just a very small, brief encounter that could have been brushed aside.

• 13:57Speaker 00

  And all of, you know, the paperwork that you're getting for a given case, it's hard to find just like those three seconds where the officer is saying, hey, just tell him to accept. But because of having a tool that made it easy to navigate all this video, she was able to clip out that portion, impeach the cop during a hearing and get the case dismissed. And so I think that's just a small example of how when you give public defenders the tools to actually sift through this footage, that's when you get true accountability. That's when you get that true transparency in the system.

• 14:31Reid Hoffman?

  What made you decide to do this as a, as a company versus, for example, a 501(c)(3)?

• 14:38Speaker 00

  There's a lot of things that I could say about this, but I guess what I'll start off with is the right to a strong and well-resourced public defender system, that's a constitutional right. That's a mandate that the state needs to be taking on and needs to be investing in. And so we're not selling to legal aid organizations that are doing this as kind of, you know, out of their own goodwill. This is a responsibility of the state as part of the democracy that we live in. And unfortunately, we live in a country where

• 15:17Speaker 00

  In many jurisdictions, the police department's overtime budget is greater than the entire budget of the public defender's office. And what that contributes to is an environment where technology vendors, they look at the criminal justice space, they look where there's opportunity. They realize there's a lot of budget for tech and innovation on police and prosecutions side of things. And so that's where a lot of the cool new technology ends up being built. We don't think that's how the system should operate. We think public defenders should have, you know, a line item in their budget to be able to invest in high-tech solutions, to be able to invest in sustainable innovation.

• 15:59Speaker 00

  And for us, it was definitely an uphill battle because hardly any of the public defender agencies that we initially started selling to had a line item for video analysis tools. By and large, they just invested in case management things and Microsoft Office. And so I think the case that we've been trying to make is, no, we should go to the county board legislature. We should go to the state legislature and tell them that this is something that you need to be able to uphold that constitutional right, and this is something that's going to make you a lot more efficient. And so I think from the beginning, one thing that we realized is that if we give Justice

• 16:40Speaker 00

  Tech as a free resource, it's simply not going to be successful. Because it takes a lot of hard work and effort and, you know, just like thought and intentionality on the public defender's side to be able to roll this out efficiently. And so when they go through all of that process of getting that budget approved, that's when, you know, Justice Tech becomes a core part of their workflow that we can then double down on. And then from our side, It's one thing to kind of just build a fun school project and give it away for free. But if we want to continue hiring the best engineers, if we want to continue incorporating the latest technology in Justice Text and expanding more file formats on our platform

• 17:25Speaker 00

  and providing that constant support, we need to be able to fund that ourselves. And so we can either spend a good chunk of our time, you know, going out and seeking charitable grant funding, which is a fine option, or we can go directly to the source, directly to the customers that we're trying to serve and make sure that this is something that they're actually willing to pay for. So that's kind of how we've thought about it and our positioning in this space.

• 17:52Reid Hoffman?

  Well, it makes sense, especially to have the kind of the connection of the customers, the fact that they have to, you know, they have to adopt their work process as part of it. So engaging in a customer relationship versus a, you know, what may be kind of tools that they never invest in as part of doing it. Isn't one of the things that that then adds to your challenge that, you know, there's limited budgets, it's, you know, how do people, you know, get persuaded this is something to, to kind of pay attention to? Like, what does the sales cycle look like? Because it's almost like you're creating a new market here.

• 18:30Speaker 00

  Yeah, and that's been a really fulfilling part of this journey, honestly, because we truly are sitting down with a lot of public defense leaders, very well-respected folks who are then coming to us and asking for advice on, hey, how can we pitch this technology? I have a, you know, county board meeting coming up. Many times we are co-creating narratives with them, we are putting together, you know, entire data sheets, fact sheets to help them make that case. And over the past year, year and a half, We've noticed that public defenders have been able to meet this case successfully. And that's been really exciting to see, actually.

• 19:15Speaker 00

  When it comes to navigating procurement, there's so many different options that you can go through. Sometimes, you know, there's like a competitive bidding process. Sometimes you might be below the threshold of what needs to be put out for a competitive bid. Sometimes, you know, if you have a very differentiated solution, you can go through sole source procurement. So for us, we've had a huge learning curve of understanding what's the best way to partner with these government agencies and navigate those sales cycles. But honestly, I think that's become like one of our core competencies at this point. And we really started off as just like two college students knocking on a lot of folks' doors.

• 19:58Speaker 00

  And people have seen us over the years just continue to keep at it and continue to grow the scale of our impact. I was at a conference with a lot of public defense leaders earlier this year. And so many of them had a memory of getting a cold email from me when I was like a senior in college. And it was, it was just cool because I feel like they recognize our genuine interest and motivation in serving them. And for that reason, you know, they've been willing to kind of give us tips and tricks on how to navigate kind of the, the sales and procurement side of things as well.

• 20:31Aria Finger?

  So I love the idea that you're not going to let the government, the federal government, the local government, whatever, like pass the buck on this responsibility. You're like, no, no, no, this cannot be legal aid. This is in the constitution, like you guys have to do this and you have to pay for it. So it is meaningful to you. Can you give us an idea of who, who and where are you serving? So we have an idea of your scale. And then obviously you were created to help public defenders. And you said you want to get to sort of truth and accountability. But do prosecutors use you? Is that okay? Like, how do you think of the flip side of the coin?

• 21:05Speaker 00

  Yeah, we're working with around 60 individual public defender agencies at the moment. And that includes entire statewide systems like Tennessee and Massachusetts, where we rolled out earlier this year. Major cities like Portland, Houston, Austin, which is where I am right now. But also small, like rural, public defender agencies with maybe five attorneys. And it's actually been a challenge to figure out, us as a technology provider, what does that customer support, that customer success function look like for us? When we're serving a massive system with a lot of resources, a training director

• 21:49Speaker 00

  versus a much smaller office that may not have a lot of in-house technical support. And so that's been really interesting because if you're trying to work with public defenders in America, you have to recognize that the delivery model is very different, county by county, state by state. I think one thing that we tell all of our partners is that you're not just getting a piece of software. We will sit down with you. We will do office hours with any attorney who wants it. We will train them. We will come visit you in person. And that's just been such a critical part of being able to serve this community, given how overwhelmed and overworked a lot of them are.

• 22:31Speaker 00

  And then on the other side of things, I think we have been very focused on public defenders and the gent defense from day one. I think it's core to our ethos, just recognizing that there are a lot of folks who are building cool new tech solutions for prosecutors, for police. But this is a very, very serious and needed gap that we can focus on. We have branched out to private criminal defense attorneys. And so they may be representing individuals who do have the means to hire a lawyer. And so I do think there's adjacent communities, adjacent markets that we have already started to serve, where there's a lot of opportunity for growth and expansion.

• 23:16Speaker 00

  But I do think we have stayed true to our mission of focusing on the defense in our journey up until now.

• 23:28Aria Finger?

  We'll be back right after this break.

• 23:36Aria Finger?

  And now, back to the show.

• 23:41Reid Hoffman?

  So let's bring in a clip from a public defender that that works with you.

• 23:47Speaker 03

  Being able to upload almost all forms of audio, whether it's a court hearing or body worn camera footage, radio traffic or witness interviews, and then being able to search the accompanying text, make clips, add subtitles, and download excerpts is incredibly helpful. So for example, I recently had a case with about 1,000 body worn camera clips. And there's no question that we were using justice text. At least, we were the most prepared attorneys in the room, having been able to evaluate each and every minute of the digital evidence to find bits and pieces that furthered our defense and highlighted problems with the state's case. In that particular case, the state would call a witness without any prior notice.

• 24:28Speaker 03

  We could use justice text to access that officer's body worn camera in real time, scan through the transcript while the officer was testifying, and gather impeachment material.

• 24:37Reid Hoffman?

  That's a clear, awesome use of justice text. What are kind of features that are being asked for? Or what are the kind of additional products to to elevate the game here and make sure that, you know, folks who are using, you know, public and, you know, eventually private defenders too, to have a, you know, kind of a robust defense?

• 25:03Speaker 00

  One really fascinating request that has only come up over the last couple of months is the ability to actually upload a police report and then for us to use kind of our AI assistant, our generative AI capabilities to automatically flag potential inconsistencies between what was captured on footage and what was written in that police report. We've also heard folks ask us about, you know, I'd love to be able to just like find every instance in which this red car showed up on video. We've also not done as much of the visual analysis component, but I think it's something that more and more people are asking for. So yeah, those two areas we're really excited about.

• 25:46Aria Finger?

  And like you said, one of the reasons you want to be a for-profit and get the best engineers and have the latest technology is so you can always be updating your tools, whether that's because new technology is arising, whether that's because new use cases are coming up, you're talking to new customers. Would you mind running us through a demo of a new feature of Justice Text?

• 26:05Speaker 00

  Yeah, 100%. So essentially, what I'm showing you right now is the Justice Text platform, where if you're an attorney, you can go ahead and upload your audio and video discovery. I have uploaded a dash cam video in a courtroom proceeding to this case folder. And one of our most recent features on the platform is an AI assistant that we call Miranda AI. What Miranda AI can do is take an entire case folder's worth of discovery, whether that's dozens of phone calls or dash cam videos. It will scan across all of them, generate a five to 10 sentence summary of the main topics of conversation. And then you can actually interact with Miranda AI

• 26:50Speaker 00

  and ask it high-level, natural language questions that help you discover all of that or understand that evidence better. So I can ask it to generate a timeline of key events. This is one of the core functions of what a defense attorney is trying to accomplish, especially if they're getting body cam videos from multiple different officer perspectives, and they're just trying to piece together who said what at what time. Having access to this tool can be really powerful. We had an attorney out in Oregon who was able to use Miranda AI to build a high-level timeline of multiple different officer perspectives.

• 27:31Speaker 00

  And he basically said that the response he got in a matter of 30 seconds was the same thing that he had compiled over hours and hours of poring over all of the footage. And so another thing that you can ask this tool is, you know, this is a DUI stop. Is there any evidence of intoxication?

• 27:54Speaker 00

  And what the tool will do is when it generates its responses, it will reference a very specific file and timestamp. And so if an attorney is using Justice Text and they want to cross-reference any of the responses that this assistant is giving them, you can do that very easily just by hovering over that response. It's something that we highly emphasize when we train new attorneys on Justice Text as well. Because like with any AI tool, there are possibilities of, you know, information that may be left out. So yeah, if I wanted to take a look at this instance where they're,

• 28:37Speaker 00

  you know, asking the individual to track a red light, I can just click on that and Justice Text will navigate me to that point in that particular dashcam video.

• 28:49Aria Finger?

  Thank you so much. This is fantastic. And I think obviously with AI, a lot of the worries about what's going to happen to jobs. And I share those concerns. And I am so worried about what's going to happen over the next, you know, two, three, five, 10 years. But I think this gives a great example of AI that is helping folks do their jobs. And they're not taking away anyone's job because these are jobs that just weren't getting done. This is stuff that underfunded public defenders, departments, underfunded, you know, legal, legal folks who are trying to do the right thing. They just didn't have the time to do this. And so without the technology, it just wouldn't happen.

• 29:27Speaker 00

  No, I appreciate that point because the work that draws public defenders into choosing this line of work is wanting to help people in their community, is wanting to, you know, provide alternatives to our incarceration, to understand their perspectives, understand their limitations. It's really that like human-centered work.

• 29:54Speaker 00

  A lot of public defenders love being able to show up as, you know, friends and confidants to the clients that they're serving. And unfortunately, just the grueling nature of their work takes away from a lot of that fulfillment many times. These attorneys are getting burnt out. I hear from a lot of offices that are suffering from retention issues. And so the way we think about it is how can technology help them double down and focus on the work that initially drew them to this profession in the first place?

• 30:25Reid Hoffman?

  So one of the things about entrepreneurs who go into this is you now have a, you know, kind of good understanding of the market, the ecosystem, the network, what's going on. What do you think are the key elements of kind of technological support? To try to, you know, best balance the, the, the, as it were, the, the, the under-resourced within the judicial system?

• 30:54Speaker 00

  The reason why the intersection between government and technology is so exciting is because low-income communities in this country, communities of color disproportionately rely on the government to operate efficiently. Whether that's for subsidized healthcare services, whether that's for, you know, social security benefits, whether that's food stamps, legal services, what have you. And

• 31:24Speaker 00

  for the longest time, we simply have not delivered a reasonable user experience to these communities.

• 31:34Speaker 00

  We've been okay with inefficient systems, with government forms that don't work, or having to call, you know, some hotline that doesn't pick up 10 days in a row before you can actually get a resolution. And, you know, if Things like our DoorDash app works so seamlessly and so efficiently. Like, we can strive for better within our government services. And that's why I'm super excited about the potential for technology to be incorporated by different agencies to solve their specific needs. As an outsider, I would have never known public defenders are struggling with body cams. So I can't even tell you the specifics of how this technology can be used or applied.

• 32:19Speaker 00

  But I think it's kind of a culture shift around our government agencies thinking about what are the unnecessary inefficiencies in how we deliver our services and how can we address them. There's so much low-hanging fruit. You don't even need generative AI. Sometimes it's just a basic type form or a Calendly integration that could truly cut down on hours and hours of work that someone is currently manually poring over. So you really don't even need to use the most high-tech fancy technology to have a really big impact in this space.

• 32:55Aria Finger?

  So going on that same theme of local governments being able to utilize technology, to your point, to be able to provide better government services, I do think actually this is like a key. Like, people would believe more in democracy and government if our government worked for them, you know, even if it was as simple as filing taxes or whatever it might be. But do you think that these local governments are starting to shift their perspective on AI? Or when you mention AI, is that a positive, a negative? Are people nervous? Does that make them excited? Like, how does that interplay with, you know, all of the new media around AI and its capabilities, both for good and bad?

• 33:33Speaker 00

  That's a really good question because I have seen such a stark shift in how these governments are thinking about AI over the past six months or so.

• 33:45Speaker 00

  We are getting AI questionnaires from every county, every state. They have entire written policies that they want our responses on. As a technology vendor, how are you storing your data? How are you securing it? What are you training? Or what are you using to to train these models? Are you using our data to train these models? And honestly, I think it's an incredible step in the right direction because the answer can't be we're not going to use AI. Or we're not going to touch this technology. Let's ban it outright. I think local governments are realizing that there's a lot of potential here. But they also need to be concerned with the potential risks. Like, a lot of this data is not public information.

• 34:28Speaker 00

  And if it's used to train, you know, a lot of these public models, that could seriously harm a lot of folks. And so I've personally been excited to see this. It creates more work for us, honestly. Like, we have a lot more that we need to do to, you know, get Justice Text vetted and get it over the line when we're going through that sales and procurement process. But I think this is the start of a really cool era in which I think governments want to be on the cutting edge of this technology and do so in a responsible manner.

• 35:01Reid Hoffman?

  Have you thought about what are the other things that could be part of the, as it were, public infrastructure? Like, does it make sense to have software API inclusions in body cams or other kinds of things that would help the, you know, the kind of the justice process? And and what kinds of things would those be?

• 35:25Speaker 00

  Definitely. I think one of the big challenges in operating within the criminal justice space has been you have a lot of

• 35:36Speaker 00

  Different players who are creating proprietary technology that isn't interoperable with, you know, their competitor's system. And it goes all the way from jail call providers all the way to who's recording the courtroom proceedings to, of course, the body cams. And I think it really is gonna have to be local governments that push a lot of these vendors to create APIs and a greater ability for other systems to interact with their own. Because honestly, in building Justice Tax, that's been a huge hurdle that we've been encountering and actually made really great progress towards.

• 36:21Speaker 00

  We get a lot of proprietary file formats. Sometimes we go to a county and 80% of the footage that they're receiving requires a special file player to be able to view. And attorneys are spending more time trying to get, you know, that file to play than they are even reviewing it in some cases. And so we've reverse engineered a lot of these proprietary file formats so that they upload and operate seamlessly on Justice Tax. But that's not necessarily what we should have to do. These private vendors shouldn't have so much say and authority in those types of decisions. There's also risks. We definitely don't want to be contributing to an environment

• 37:04Speaker 00

  where every action anyone makes while walking on a public street can be easily surveilled and used for analyses by their local government that they may not be consenting to. And so when it comes to the use of that hardware and software, I think that's something that we all have to be wary of. But yeah, I definitely think there's potentials for more integration.

• 37:30Reid Hoffman?

  So one question I love to ask entrepreneurs is, if you had a phone line that goes through time and you call your younger self to, not with, oh, and the world looks like this, or, you know, talk to Sue but don't talk to Bob, but like a decisioning learning algorithm, like what to do more of, what to do less of. What would you tell your younger self?

• 37:55Speaker 00

  I think for me, what I would tell my younger self is, build up that skill of delegation because it is genuinely so. Cool. And it's so beautiful when you bring on a new team member, you show that trust in them, you give them a responsibility and they're able to execute it better than you ever could have imagined. And the best leaders that I see are the ones who are able to surround themselves with other stellar individuals and just inspire them to do big things. And it's something that I continue to strive for and am actively working on. But I think this next stage of scaling Justice Text is really going to be about investing

• 38:37Speaker 00

  in the people that I work with and just, you know, thinking about their leadership journey and growth.

• 38:45Aria Finger?

  So on that note, give us a little bit of a preview and you can answer this however you like. Like, what is the future for Justice Text? Are there adjacent fields? Do you want to be in all 50 states? Do you have a five-year vision? Like, just give us a little bit about the future for you guys and what you're shooting for.

• 39:02Speaker 00

  We are in a time of really rapid growth right now. We are close to finalizing two additional statewide partnerships in addition to the two states that we're already fully deployed in. We have tons of like growing interest in states where there's, you know, increasing pools of funding for discovery and body cam analysis and all sorts of stuff. So I think we're at a really, really exciting point in our journey where I'm very, very motivated to continue pushing forward and continuing to get our product into the hands of more public defenders. We recently partnered with this really incredible organization called

• 39:44Speaker 00

  Zellis. They're focused on media and storytelling to combat outdated narratives around criminal justice in the country and to push forward like a more progressive view of what public safety looks like. And we partnered with them to make a video about our work in Kentucky with their public defenders. And so we did a road trip around the state, met a lot of our attorneys, told their stories, thought that was beautiful, that was fantastic. So I'm excited to share that and I'm excited to do more things like that. And then lastly, on the product side, it really is what I was talking about earlier. You know, just the advancements in generative AI. It seems like every week there's something new going on that we need to keep an eye out for.

• 40:26Speaker 00

  And so whether that's using LLMs to analyze more of that written discovery, analyze police reports, or start doing more of that visual analysis. I think that's where a lot of Leslie, who's my co-founder, where he's going to be focused over the next couple of months.

• 40:51Reid Hoffman?

  So let's do rapid fire. Is there a movie, song, or book that fills you with optimism for the future?

• 40:58Speaker 00

  There is this one movie, this Indian movie called Queen, which honestly is probably one of my favorites. It's about a young woman who was essentially left at the altar and decided to go on her honeymoon to Europe by herself. And she had never left her small village in India before then. And, you know, shows up in Paris, not really knowing the language, not understanding a lot of the cultural nuances. And so there's a lot of miscommunication that takes place, a lot of embarrassing moments. But she pushes through that and she built a life for herself.

• 41:40Speaker 00

  She is able to de-center this idea that marriage is the one and true source of value in her life. And honestly, it's just like such a cool and beautiful story. So me personally, as a young woman who comes from a culture where women aren't always able to travel the world, start their own companies, and live life on their own terms. I think watching that movie means a lot and fills me with optimism as well.

• 42:11Aria Finger?

  There's a lot of things to fix in the world, but it's also pretty cool to think about how far we've come as a gender, as a species, as all of the things. Rapid fire number two. What is a question that you wish people would ask you more often?

• 42:26Speaker 00

  I love talking about my experiences solo traveling, living abroad. I spent a good chunk of time over the past two years living in Latin America. So did my co-founder. And you know, when you show up in a country where you don't know a single soul, where you don't know the language and you still have to fend for yourself, you learn a lot about yourself.

• 42:45Reid Hoffman?

  So where do you see progress or momentum outside of your industry that inspires you?

• 42:52Speaker 00

  I am really inspired by how multicultural a lot of the stories that are being told in the media are now. If you go to Netflix, Amazon, Hulu, it's really cool to see a lot of just like niche stories, niche communities, having entire movies or entire TV shows made about them and their lived experiences. Especially when you live in a country like the United States that is so deeply multicultural. And unfortunately, also to this day, pretty segregated. There's no way that you're going to have the lived experience of every other American.

• 43:37Speaker 00

  So yeah, I'm excited to see more and more untold stories come to life on on screen.

• 43:43Aria Finger?

  Can you leave us with a final thought on what you think is possible in the next 15 years? If everything breaks humanity's way and what's the first step to get there?

• 43:54Speaker 00

  One thing that I think a lot about is how a lot of the power of AI and automation has been restricted to the world of software up until now. Like, how do you take enormous amounts of data and synthesize it efficiently and quickly? But there's a lot of very intelligent people who are working on How do you use, you know, physical objects, whether that's cars or robots, little vacuum cleaners to automate parts of, you know, our day-to-day physical lives? And I think about that in the context of the burden of domestic labor, especially on women.

• 44:36Speaker 00

  So many women are now breadwinners in their family or, you know, pursuing these incredibly fulfilling careers and still grappling with the incredible burden of managing the household, of taking care of all these kids and all their activities. How can that next wave of AI, of automation, help alleviate some of that burden so that families can focus on the things that matter the most, which is spending time with one another? I think this has implications on elder care as well. I think I hear from so many friends who have grandparents that, you know, are struggling

• 45:17Speaker 00

  with this new era in their life where they don't have a lot of people that can talk to them. They're not physically able to do the things that bring them joy. And unfortunately, you know, working parents are already incredibly overwhelmed. So I don't really have any answers, but I'm very, very excited to see how a lot of innovation kind of on the hardware side of things can help to address some of those burdens.

• 45:48Reid Hoffman?

  Possible is produced by Wondra Media Network. It's hosted by Ari Finger and me, Reid Hoffman. Our showrunner is Sean Young. Possible is produced by Katie Sanders, Edie Allard, Sarah Shleid, Adrian Bain, and Paloma Moreno Jimenez. Jenny Kaplan is our executive producer and editor.

• 46:07Aria Finger?

  Special thanks to Surya Yalamanshili, Sayida Sapieva, Ian Alice, Greg Beato, Parth Patil, and Ben Ralas. And a big thanks to Jonathan Irvin and Little Monster Media Company.