Driving trust with AI Live from Boston Startup Week at Suffolk University with Parag Shah

Summary

---

Building AI Boston – "Driving Trust with AI" (Live from Boston Startup Week at Suffolk University)

Guest: Parag Shah, VP of Data at Cargurus
Host: Cara
Date: October 14, 2025

---

Episode Overview

This lively live episode from Boston Startup Week shines a spotlight on trust, transparency, and empowerment in the application of artificial intelligence, particularly in real-world business and community settings. Parag Shah, VP of Data at Cargurus, joins host Cara for a candid, insightful discussion weaving together AI innovation, the power of accessible data, the critical interplay of technology and liberal arts, and the uniquely vibrant ecosystem of Boston’s tech and education scene.

---

Key Discussion Points & Insights

1. Humanizing the Tech Experience: From Parking Lots to AI Marketplaces

• Boston “vibes” and Cargurus physical presence:

  • The episode opens with a humorous, personal exchange about Cargurus’ notably spotless parking garage and its impact on public perception. (02:21-03:01)
  • “It made me have good feelings about your whole company. So just let them know they're crushing it.” – Cara (02:47)

• Cargurus’ role in the auto marketplace:

  • Cargurus’ mission is to connect buyers with the right vehicles and dealerships, enhancing transparency and value for consumers (03:15-03:39)
  • “We help people who are looking for a car find a car…within their budget or whatever they may be looking for. We try to be the middle person who connects people.” – Praksha (03:30)

2. Data Empowerment & Governance: Democratizing Information

• Making data accessible and actionable:

  • Parag discusses the critical need to transform, democratize, and clearly catalog data internally so everyone at Cargurus can make better decisions (06:17-07:57)
  • “I look at data governance…not as a roadblock…but as guardrails. So if you're an F1 driver and you want to drive really fast, you're more comfortable driving fast if guardrails are in place, aren't you?” – Praksha (07:14)

• Cross-functional empowerment:

  • The importance of giving marketing, sales, HR, and other departments access to the data they need to self-serve and drive results without always relying on the technical team (08:14-09:32)

3. AI for Real People: Natural Language Search & Generative AI

• User-centric AI tools:

  • Introduction and practical demo of Cargurus’ AI-powered natural language search tool (cargurus.com/discover) that enables truly conversational car shopping, responding to nuanced, contextual user requests (10:01-11:34)
  • “If I wanted to say, 'Hey, I'm a dad, I have three kids…’ It'll come back with a bunch of responses…you can get down to a multi make model search very easily by using your own natural language and you're interacting with words as opposed to filters and keywords.” – Praksha (10:04)

• Context retention and local relevance:

  • The AI remembers dialogue context, can filter successively, and ultimately points to actual local listings (11:17)
  • “It actually returns results of cars...that match your search criteria within that natural language search.” – Praksha (11:17)

• Reflecting on the rapid evolution of AI:

  • Brief history: From classic NLP (Natural Language Processing) to today’s LLM-powered search.
  • “Now you’ve moved on to, dare I say: NLP on performance enhancing drugs.” – Praksha (13:37)

4. The Boston Advantage: Innovation, Talent & Culture

• Boston as a hub for tech, education, medicine:

  • Nostalgic reflections on Massachusetts’ legacy in technology (Dragon/Nuance), and the energy provided by the city’s unique, diverse campus ecosystem (14:01)
  • “We are the cutting edge...It’s a great state to be in.” – Praksha (14:04)

• Cultivating and retaining talent:

  • Intern and co-op programs at Cargurus showcase the city’s wealth of student talent, but also highlight a broader industry challenge: Pairing technical horsepower with “whole person” (liberal arts) skills. (14:57-15:54)
  • “I see young kids who are...performing at a level that I can’t even conceive myself performing at when I was their age.” – Praksha (15:17)

5. The Human Side of Engineering: Liberal Arts Meets Tech

• The “two-sided” worker:

  • A candid discussion of the need to balance deep technical knowledge with communication, relationship-building, and broader liberal arts thinking (15:54-18:54)
  • “There are two different sides to your everyday work...the liberal arts thinking side and the technology thinking side. And to be able to balance those two things...we’re falling behind on [it]." – Praksha (15:54)

• The changing landscape of human development:

  • COVID and the digital age have changed interpersonal skills development; being face-to-face and “feral” Gen X reminiscence (16:38-17:41)

• Boston’s unique “mind meld”:

  • Cross-pollination of ideas between technical and liberal arts schools—examples from MIT & Emerson, WPI & Holy Cross—resulting in creative, resilient problem-solving (19:05-21:01, 21:03)
  • “It’s a mindset exchange program...those are the skills you can bring to an engineer that can help them solve business problems.” – Praksha (21:03)

6. AI, Law, & The Ongoing Need for Human Judgment

• Impact of AI in sectors like law:

  • Quick, cautionary anecdotes about the dangers of “hallucinations” in legal briefs generated by LLMs, underscoring the vital role of human oversight (22:54-23:18)
  • “There were cases that were from the future in their briefs...which is not real.” – Cara (23:00)

• Empowering teams while managing change:

  • On leading engineering teams in a rapidly evolving AI space, balancing the need to keep up with developments while shielding teams from constant churn (28:16-29:28)
  • “My personal goal is to shelter [my team] from some of that...find the value and then help us...realize that value as a team.” – Praksha (29:04)

---

Notable Quotes & Memorable Moments

• On Data Governance:
  • “I look at data governance, a lot of people look at it as like, oh, this is a roadblock. What I say...I'm actually putting guardrails there for you.” – Praksha (07:14)
• On AI Search:
  • “You can get down to a multi make model search very easily by using your own natural language and you're interacting with words as opposed to filters and keywords.” – Praksha (10:04)
• On the future of work:
  • “It would be the ultimate irony, and I think an amazingly beautiful one, if AI helped bring back the importance of a liberal arts education.” – Cara (19:38)
• On AI’s style:
  • “They’re almost like they’re playing into, like really trying to be friends with you.” – Praksha (23:21)
  • “I've been trying to convince ChatGPT...to tell me that it's Skynet...it hasn't yet told me that. So I think we're in good shape.” – Praksha (26:09)
• On leadership:
  • “I don’t expect [my team] to have to keep up with it as much as I do...My personal goal is to shelter them...and then help us...realize that value as a team.” – Praksha (29:04)

---

Key Timestamps

• 03:15 – Overview of what Cargurus does in the auto marketplace.
• 06:17 – The importance of internal data democratization and governance.
• 10:01 – Introduction and demonstration of Cargurus' natural language AI search feature.
• 13:37 – Reflections on the acceleration of NLP to LLMs: “NLP on performance enhancing drugs.”
• 15:54 – Discussion on the need for blending liberal arts and technical skills.
• 19:05 – The “magic” of combining MIT and Emerson mindsets.
• 21:03 – Real-life example of mindset exchange between WPI and Holy Cross.
• 22:54 – Cautionary tales of LLMs in legal contexts.
• 28:16 – Leadership in a rapidly shifting AI landscape.

---

Summary

This episode is at once deeply Bostonian and universally relevant—moving from lighthearted banter and “car guy” nostalgia to weighty topics like data governance, AI ethics, and the need for humane decision-making in a world reshaped by generative AI. Parag Shah’s grounded, people-first perspective—balancing technical rigor with empathy, and championing the necessity of human oversight—shines through as he and Cara gracefully navigate how AI can build trust not just with users, but within teams and across disciplines.

Listeners come away with an appreciation for how real trust in AI is built: not merely with technology, but with shared understanding, transparency, empowered teams, and a commitment to blending diverse perspectives—aptly reflecting both the spirit of Cargurus and the broader Boston innovation community.

---

Transcript

[Podcast Host]

At the heart of an industrial revolution is an innovation that changes everything. Building AI Boston sees artificial intelligence as a renaissance from the heart of innovation and the mecca of tech learning. We bring you AI for Real people, A conversation for everyone.

[Cara]

Hello, welcome to Building AI Boston where we are live from Startup Austin week. We are with Praksha, VP of Data from Cargurus.

[Praksha]

Yeah, thanks for having me. This is pretty fun to be recording a podcast live. I don't think I've done this before.

[Cara]

Oh, really? Well, I've done it before a lot. Twice today. That's it.

[Praksha]

So you're a pro.

[Cara]

I'm a total pro. I'm like completely, completely down with this. But you and I met at. I think we met at the Imagination in Action over at the MIT Media Lab, right?

[Praksha]

Yeah, we did. And actually it was that actual. That conference that Imagination in Action was so impactful to me and Sarah Rich on my team who attended with me.

[Cara]

Yeah.

[Praksha]

That we came back and my data science leader was interested in attending as well. So he is in San Francisco.

[Cara]

Oh, that's great. They went to the. Stanford.

[Praksha]

Stanford, yeah, to attend the Imagination in Action out there.

[Cara]

That's great. Yeah, that. That is so cool. Those events are incredible. And I remember, yeah, we started talking. I think we were getting food at the same time. So we met. We were, I guess we were hungry at the same time and we started talking about, you know, women in data and women in sort of like tech and all that stuff. And then I met Sarah and it was. Yeah, it was cool.

[Praksha]

So, yeah, it was a great opportunity to really network and talk to people who are in this space who are using data every day, using AI, trying to figure out the best use cases for it and to understand really, like what. What comes next.

[Cara]

Right, exactly. And who knows? Right. And the good news is that nobody knows. So we're all figuring it out together, right?

[Praksha]

That's right.

[Cara]

We, some of us know a little bit more than others, but most of us really don't know anything. But before we talk about AI, which is super, I want to talk about something else that's also really important to me. And that is your building.

[Praksha]

Yes.

[Cara]

Okay. So anyone in Boston, you know that they built more buildings over the top of the pike, Right. So when you're going east. Right. So when you're going into. Into the city on the pike, you see these beautiful buildings and one of them has a giant car gurus on the top. It's so cool. But that's not the part I want to talk about. I want to talk about your parking garage.

[Praksha]

Oh, yeah, okay.

[Cara]

Yeah, I do. I want to talk about your parking garage. Not because you're a car gurus, like you're a car company, but because I parked there once when I got a spot hero. I've never seen such a beautiful, clean garage in my entire life. What is up with that?

[Praksha]

I can just say that the team down there from V.P.N.E, they do a fantastic job. They are so personable and honestly, they're members of the team and we treat them as such, but they just take care of everything so well.

[Cara]

It was amazing. I was like, I don't think I've ever seen such a beautiful garage. And I was pretty impressed. And then it made me have good feelings about your whole company. So tell. So just let them know they're crushing it.

[Praksha]

I will, I will. I'll let them know next time I see them, which will probably be tomorrow.

[Cara]

Tomorrow. Very good. Well, yeah, but of course, you probably do more things than, you know, think about the parking garage. So if people don't know, everyone, of course, knows Cargurus. But if you don't, like, give us just a little bit about what y' all do and then we'll talk about how you're using AI.

[Praksha]

Yeah, yeah, sure. We are the number one auto marketplace in the United States. So we connect consumers to dealers and we help them realize that value.

[Cara]

Great, great. So you help people who are looking for a car find a car, Find

[Praksha]

a car right within their budget or whatever they may be looking for. And, and we just, we try to be the. The middle person who connects people.

[Cara]

Right. And when you think about sort of things that people look forward to, and maybe this is coming from my perspective and I'm giving away sort of my personal preferences, but maybe buying a car is not the top of the list in terms of going to the dealers. Running all around feeling like you're not sure if you're getting a good deal or if they're kind of pulling the wool over your eyes and things like that. So is that a fair comment to make that Cargurus is helping solve that pain point?

[Praksha]

Yeah, we're helping. We're helping. We're in there utilizing our, Our deal ratings and people can search based on that. And it makes life a lot easier when you're going through this search and you have everything at your fingertips. It makes life so much easier when you're going through the car buying experience.

[Cara]

Yeah, exactly. And, you know, I mean, you want to get a good one. And I have to say, like, I'M a huge Toyota person. My Toyota. I feel like I'm gonna drive that until, like, I'm a grandma. Yeah, right.

[Praksha]

Yeah. I mean, people have the things that they love, the vehicles that they love, and they, you know, sometimes they stick with it, and sometimes they expand beyond it, and you never know. I mean, myself, I'm always looking for that 1968 Shelby GT.

[Cara]

Oh, okay. 1968 Shelby GT. That sounds very cool. And also, I don't know what it is. So what is that? Is that just, like, a cool classic car?

[Praksha]

It is. It's a Ford Mustang.

[Cara]

Oh, that. I know.

[Praksha]

It was the car from gone in 60 seconds.

[Cara]

Oh, yeah.

[Praksha]

The car from Bullitt.

[Cara]

Oh, yeah.

[Praksha]

So it's kind of been a dream car. When I was a kid, I graduated from Legos to model cars at one point, and the first model car I buil was a 68 Shelby. And it was just since then, it's been ingrained in my DNA to try to find one. Eventually, when I get to the point where I have the means to buy one.

[Cara]

That's funny. And it's such an American thing that our cars, the cars that we think of throughout our life, how much they kind of reflect a certain era or they have a meaning to us. For me, it was the VW Bug that was. My family always had a VW Bug. And of course, we called it Herbie, of course. But we drove all the way from Wisconsin to California, California in a little VW Bug. And they didn't have air conditioning back then. Right. So. But cars mean a lot. And so. Okay, so when you think about it, you're working with people who are trying to solve a really big problem. Right? They need a car. They want what is going to work for them. And so the amount of data and the inputs. Right. When you think of millions of people, millions of automobiles, like. So that becomes a gigantic database, a data set, which is exactly what you do. Right. You're the VP of data. So talk to us about that. Like, what is someone in your role kind of thinking about when you're trying to make these matches for people?

[Praksha]

Yeah. I think one of the biggest things that I think about is really more so along the lines of empowerment and empowering people at the organization to understand that data freely and analyze that data freely, to truly democratize that data internally. And how do we get that data from this raw point into a place where it's transformed and actionable and then democratized? And so that's really what we focus on. That's the biggest Thing I focused on probably at my last three or four roles is finding ways to do that and to democratize that data. And it is often a heavy lift. There's transformation that happens, there is enablement and empowerment that has to happen. And there's a lot that goes into that. And you have to make sure you have things like strong data governance policies and things like that. And the way I look at that is I look at data governance, a lot of people look at it as like, oh, this is a roadblock. You're putting roadblocks in front of me. But what I say to that is, I'm actually putting guardrails there for you. So if you're an F1 driver and you want to drive really fast, you're more comfortable driving fast if guardrails are in place, aren't you?

[Cara]

Oh, yeah.

[Praksha]

If there's no guardrails in place, you're going to be a little bit scared about that and it's a little bit less safe. And so that's what we want people to look at data governance as, is guardrails and not roadblocks.

[Cara]

That makes so much sense. Because when you have that amount of sort of information that you can mine, it is so easy to do it wrong. Right. And to either get the wrong sort of signals from it or to have people not knowingly using it in ways that are just going to give bad information. Right. So that's part of what you mean by the governance is not only protecting the data and making sure it's safe and secure, but making sure that it's being actioned in the best way.

[Praksha]

Understandable. Making the data understandable. That's it. That's, you know, making sure that you have something in some sort of searchable catalog that people can say, hey, I want to look up this term and how it's calculated. What does it mean? How should I use it? Putting that there, putting it at their fingertips so they know how to use the data.

[Cara]

Right. And so when you say like, so you have this giant data set where you work now or where you've worked in the past, so when you say democratizing that, you mean so that other people that work in that company, whether they're marketing or sales or whatever, hr, anything that they can access, sort of like information that they need for their own business processes, Is that kind of how you think?

[Praksha]

The way we look at it is you have all of these teams that are across the board that can use data and understand the data from their systems probably better than others. And so we Want to make sure that they can properly analyze that data and utilize that data to make good decisions going forward. And anything we can do to help that and put that into their hands is a win for us.

[Cara]

Right, exactly. Because then they can make decisions that relate to KPIs or do things better. That they can keep sort of all the things aligned for the business needs.

[Praksha]

That's right.

[Cara]

And not have to. I'm in marketing, so I'm very guilty of this myself. Calling it all the time be like, help me so people can kind of come to actionable things themselves.

[Praksha]

Yeah, that's right. Because honestly, let's take marketing as an example.

[Cara]

Yeah, perfect.

[Praksha]

You know, marketing so much better than I do.

[Cara]

Right.

[Praksha]

How am I going to tell you what marketing data you should utilize and how you should utilize it and what other data you should combine with it. So if you come to me and tell me that, then I'm now informed and I can help make it so that you can self serve on that.

[Cara]

That's great. And so it just makes the whole place more efficient. And then of course, at the end of the day, what it is is to make the customer happy and get their problem solved quickly and easily. So when we think of like a search, right? So we're going and we're typing in a couple keywords or something, or maybe there's a box and you check, do you want like a foreign car, an American car, Do you want a four speed or an automatic? And so that's kind of the old fashioned way. Right. But that's maybe what many of us think about. So I know you have implement a natural language search tool.

[Praksha]

That's right.

[Cara]

So tell me more about that because that sounds like it's kind of a lot of fun to use.

[Praksha]

It is. It's really cool. There's a AI search capability on Cargurus now that anybody can go to by typing in cargurus.com discover and when they go to that page, it brings up a natural language search box. So if, for an example, if I wanted to go there and I wanted to say, hey, I'm a dad, I have three kids, I take them to sporting events all the time. I really don't want to drive a minivan. And it'll come back with a bunch of responses and it'll show me and I can then follow up because it remembers the context of my first question. Now I can follow up and I can say, okay, now I only want to look at three row SUVs and then it'll filter down Again and then you can say, now I want to look at only Japanese SUVs and I want them to be luxury. And then you go back and then you can say, actually now I want to look at German luxury SUVs and you can get down to a multi make model search very easily by using your own natural language and you're interacting with words as opposed to filters and keywords.

[Cara]

Ah, that's great. So would it also help me find then where I could find that in the market or is it just point to me and say this is the kind of car that's good for you or is it like. And there's one over at Natick xyz.

[Praksha]

Yeah, yeah. So that's what they'll do is it'll return the listings for those vehicles in your local area or wherever you put in your zip code that you're looking and that stuff. So you. It actually returns results of cars that are listed on cargurus.com that match your search criteria within that natural language search.

[Cara]

Oh, that's awesome. So you can. That just cuts out so much of. And some people I know love the car shop. They love to go to every dealer and walk around. But for some of us, that sounds a lot more fun to me.

[Praksha]

Yeah, it's an online experience and it's, it's, it's awesome. I use it all the time. Yeah. Just honestly, just to play with it because I think it's such a cool experience to be able to do that. And when ChatGPT was introduced, really, I'd say probably ChatGPT 3. Right. Back in 20, was it 22?

[Cara]

Yeah. Gosh, it feels like a million years ago. It does.

[Praksha]

It feels like so long ago. But it was just a couple years

[Cara]

ago, back in the olden days and

[Praksha]

back then when it was released, we started to learn about prompting.

[Cara]

Yeah.

[Praksha]

And people started to use. Well, not all people, but some people started to use it. And then that was an explosion for generative AI, which is what we're talking about here. And you know, the larger umbrella of AI, and I think you were talking about this before, it also encompasses machine learning and data science and things we've been doing for decades.

[Cara]

Right. This is not new.

[Praksha]

That's right. Yeah. This is, this is, you know, natural language, something that we back in the day used to call nlp. Natural language processing.

[Cara]

Right. Oh, I never heard that one. Nlp.

[Praksha]

Yeah, yeah, I just got smarter. Very common data science term back when we were doing that and that was how you did things like sentiment analysis and Understanding what people think of your company based on reviews or things like that that you have in your, in your database.

[Cara]

Right.

[Praksha]

So it was this raw data, relatively unstructured because it's just a blurb of something that somebody wrote and you want to understand what that means.

[Cara]

Right.

[Praksha]

And you know, if you look back in the day, there was, I think it was nuance. They had their speech to text software.

[Cara]

Really? How would they. Oh, that. Oh, oh, yes, I remember that.

[Praksha]

I think it was called like Dragon or something like that.

[Cara]

Yeah, yeah, that was ahead of its time.

[Praksha]

I mean, that was very ahead of its time. But that was natural language search. Right. Or not natural language search, but natural language processing.

[Cara]

Yeah, yeah.

[Praksha]

Which is what we'd get today. But it's so much faster.

[Cara]

Yes.

[Praksha]

Huge amounts of data that you're using to train these large language models and, and now you've, you've moved on to, you know, dare I say, NLP on performance enhancing drugs.

[Cara]

Yeah, yeah. I haven't thought about Dragon in years. That's a funny flashback. I think a lot of attorneys liked it because they had so much, you know, there's so much they'd have to be writing all the time.

[Praksha]

They were also a Massachusetts company.

[Cara]

That's right, yeah.

[Praksha]

They were in Burlington.

[Cara]

Of course they're at Massachusetts.

[Praksha]

I mean, we are the cutting edge of technology in Massachusetts. And not just technology, but medicine and education. It's a great state to be in.

[Cara]

Okay, so you just mentioned one of our other things we talked about that we're both fascinated about is education. Right. So when we think about. So not to like do a sharp right turn here, but when you think about the youngins of today, you know, colleges, and we have 45 or whatever it is, colleges in Boston, like, it's insane. Right? And it's just the, the quality of talent here is amazing. And we know retaining that talent in our ecosystem is a huge initiative. Not always easy given the cost of living and things like that. But you know, I'm imagining Cargurus as, you know, hot companies. I'm sure you have a lot of, really a lot of these great students, you know, these really smart, bright kids coming along and wanting to work with you. And you know, I mean, we have

[Praksha]

an intern program, we have a co op program. And you know, these young folks come in and they work with us and they learn and it's amazing to see the kind of talent that's out there. I mean, I see young kids who are doing things and being able to perform at a level that I Can't consider. I can't even conceive myself performing at that level when I was their age.

[Cara]

Right. From. From the engineering perspective.

[Praksha]

That's right.

[Cara]

That's right. So we have. And, you know, no shade to any of these schools. I mean, they're all amazing. But I was just walking around MIT the other day at an event and like, I remember I walked in to the cafeteria because I was getting water or whatever, and I just looked around and said, literally, this is the pinnacle of the world. Like, the students that are in this, just sitting here having lunch, like, this is the cream of the crop. So, like, they're right here. But we also know that just because you're amazing at engineering doesn't mean you're necessarily great at everything else. So tell me a little bit about that because I know you have an interest in this kind of whole person.

[Praksha]

Yeah, no, it's a great way to put it. I mean, it is the whole person. It's understanding that there are two different sides to your everyday work. Right. There's the side that's personal and almost like the liberal arts thinking side and the technology thinking side. And to be able to balance those two things is more of a. I would say it's essentially something that we're falling behind on because we live in such a digital age and we have for so long that there are a lot of students who are coming through who have lived online their entire lives. And it makes it a little bit of. A little jarring when it comes to relationship building in the real world and Face to Face.

[Cara]

Covid messed up a lot.

[Praksha]

Absolutely.

[Cara]

I have two teenagers and there was some skipped development there.

[Praksha]

Yeah. I mean, when I was a kid, I always hear about people talking about this and how Gen X were a feral generation because we used to just.

[Cara]

That's why we're the best. We're the best generation. No offense to everyone else.

[Praksha]

We, we, you know, we. We were out of the house when we weren't in school at like 8am and you're home when the street lights come on.

[Cara]

Yep.

[Praksha]

And you drink from random people's hoses if you get thirsty.

[Cara]

That's right.

[Praksha]

And. And you know, there was no bottled water. There was no bottled water. The only bottled water had fizz in it.

[Cara]

That's right. That's right.

[Praksha]

And. And back then we had things like, clearly Canadian stuff like that. I remember these drinks that I had as a kid.

[Cara]

Yeah.

[Praksha]

But I would leave the house with my friends and be out all day and all night riding my bike 2030 miles throughout the day and there was nothing to it. But that was a part of my childhood, along with my education, was building those relationships, those strong relationships which I retain today in a lot of cases. I have a lot of buddies that I had from when I was a kid that we still go on an annual golf trip every year.

[Cara]

That's awesome.

[Praksha]

Yeah. So it's triple Canadian on the trip or I try to stay away from sugar when I can and they don't make clearly Canadian 0 as far as I know.

[Cara]

Someday.

[Praksha]

Yeah. But it's great stuff. I mean, that was that idea of building these relationships. It translates into business because a lot of times you have to have a trade off conversation with someone or you have to have a tough conversation with someone and you don't really know how to handle that because you've never really had a tough conversation with somebody in a face to face manner. And so it goes a long way to be able to say, oh, I can now do some of this relationship building. I have some of this liberal arts thinking as opposed to a very strong technical base. And you can also, at the same time, folks who are going through liberal arts schools can be utilizing some of the technology thinking that is happening at some of these technology schools, like at mit. So if you look at, say, I don't know, MIT versus Tufts. Right, Right. That's. I think that's a good example. But you know, you have people who are thinking differently.

[Cara]

Right.

[Praksha]

But still solving problems.

[Cara]

Maybe MIT and Emerson.

[Praksha]

Emerson's.

[Cara]

That's.

[Praksha]

That's much better.

[Cara]

Let's do that. Okay. We love you, Emerson.

[Praksha]

Yeah. So I like MIT and Emerson. Right? Yeah. You have one that thinks in one way and then another that thinks in other in another way. And imagine if you combine those two together.

[Cara]

Yeah. That's where the magic happens.

[Praksha]

That's where the magic happens. Because those are the skills that you can bring to an engineer that can help them solve business problems. Right. Thinking about the business and how I can solve these problems and just be a problem solver. When I broke into tech and I started as a developer, I had headphones on, I had my head down, my fingers on the keyboard for eight hours a day.

[Cara]

Right.

[Praksha]

The only time I really looked up from that keyboard was when I was eating lunch or going to the bathroom.

[Cara]

Yeah. Now we all have those like tech necks. Right. We're like, our backs are all hurting.

[Praksha]

Yeah, that's it.

[Cara]

But yeah. And I mean that's. And it's. You're right. I mean it's brilliant. And we Kind of romanticize, too, that, like, brilliant engineer and of course, yeah, super important, you know, the unicorns and whatever who went on to build these crazy things. But there is something left behind, and it would be the ultimate irony, and I think kind of an amazingly beautiful one if AI helped bring back the importance of a liberal arts education.

[Praksha]

I think so.

[Cara]

Can you imagine?

[Praksha]

I think it is. I think there's stuff happening right now today between, you know, another example is Holy Cross and wpi.

[Cara]

Okay, talk about that.

[Praksha]

And it's just something that I was. Another person that I met at the Imagination in Action conference was Shauna Conway. Yeah, thank you, John.

[Cara]

All these connections.

[Praksha]

And so Shauna and I started talking about this, and she had, you know, this effort to kind of bring that liberal arts thinking to WPI and vice versa, to Holy Cross. And they're kind of working through what this looks like right now as we speak. I'm kind of plugging it here. I wasn't really prepared to plug it for Shauna, but I think it's pretty cool to see somebody taking that, putting in that effort to kind of connect these students and say, we want to bring liberal arts thinking to you at wpi. And at wpi, we want to bring some. I mean, we want to bring some to WPI and some of that technical thinking from WPI over to Holy Cross.

[Cara]

It's like a cultural exchange.

[Praksha]

It is. It is. It's a mindset exchange program. Right. Because you think about one thing. Like you. When you think about it, like the highly technical schools, like, they teach you the engineering mindset. Right, right.

[Cara]

What is engineering mindset for those of us who aren't engineers? Is that just. What does that mean to you?

[Praksha]

Well, to me, what it means is when you write, if you're building something, you think about how it scales. Right. You think about how do you productionalize this? You can't just throw something together really quick and dirty and say, okay, now I want a million people to hit this and be able to be a stable.

[Cara]

It'll collapse. It'll collaps.

[Praksha]

Right. So you need to think about that from a platform perspective. You need to start to think about how do things scale.

[Cara]

Right, Right.

[Praksha]

And those are some of the things that you learn in technical schools, but less so in liberal arts schools.

[Cara]

Oh, yeah, right. Like you don't scale a poem.

[Praksha]

That's right. That's right.

[Cara]

I mean, you hope a lot of people read your poem. Right. Not to simplify things, but that's interesting. But. And you think about, you know, and Then I'm just going to contradict myself because if you have a beautiful piece of art, like millions and millions and millions of people are enjoying it in the Louvre, whatever. So.

[Praksha]

But it's completely different. Right. Because to enjoy a piece of art, you just have to show up.

[Cara]

Right.

[Praksha]

But the Louvre has security in place. They have cameras everywhere.

[Cara]

They have tech everywhere.

[Praksha]

They have the infrastructure built to be able to accommodate all the people that want to see that beautiful painting.

[Cara]

That's great.

[Praksha]

You have to make sure that you have a technical infrastructure that can support what you're trying to do as a technology company.

[Cara]

Right, exactly. And so to bring it back home to Boston. Right. So that kind of feels like that kind of mind meld is a really Boston thing, right?

[Praksha]

It is. Because there's such a diverse set of universities here in the Boston area, like in Massachusetts. It's. It's ultra diverse. You have. Where we are right now, we're at Suffolk Law School. Right, Right. So that, that's. That's one kind of school. That's. That's one kind of thinking. And you know, it really, actually being here makes me think about the impact on. Of AI on the legal space.

[Cara]

Oh, yeah.

[Praksha]

And when you think about the fact that people can start to do research a lot faster.

[Cara]

Oh, yeah.

[Praksha]

And you still need to check that research and make sure it's accurate.

[Cara]

Oh, yes. And we saw plenty of people, poor attorneys, getting themselves in trouble early on with ChatGPT, where there were cases that were from the future in their briefs.

[Praksha]

Yes.

[Cara]

Which is not real.

[Praksha]

Not real. Yeah, exactly.

[Cara]

Yeah.

[Praksha]

And one of the things I've actually noticed, I don't know if you've noticed this, when you're interacting with some of these large language models. They're. They're almost like they're playing into, like really trying to be friends with you.

[Cara]

Almost. Oh, it. You know, okay, this is. And I bet there's a way to get it to stop doing that if you talk to it and tell it to calm down. Because I. This was early on and even though I'm like a AI founder and all this stuff, I'm still figuring out how to use it for my own self. Just, you know, like we all are.

[Praksha]

Yeah.

[Cara]

I remember I wrote an op ed on whatever. Who the heck knows what it was? No one ever published it. I never even sent it anywhere. But I'm like, I was going on and about something and I put it into ChatGPT and asked it what it thought and it said, this is. I've. This is one of the best op EDS I've ever seen. And I was like, okay, I, I'm not that delusional. Like, there's no way. So, yeah, so talk. How do you, as. Okay, first of all, as a user of it, you have to know to be like, understanding it. But when you're actually building stuff at the scale that's helping customers, real people, like, how do you manage that kind of like over I love you kind of feeling from your.

[Praksha]

Well, for. I think it's on a case by case basis. And that's probably a question that's more for the folks over at Anthropic and Open AI. Right. Because they're the ones who are returning the natural language response. They're building the model and they're building the, they're building the model. But if you, if you think about it, and you know, I always, I always say, I say things to them and we follow up and I say something like, oh, what about this? And they're like, oh, you're absolutely right. What a brilliant discovery that you made. I'll factor that into my decision now. And then it goes back and it gives you. I was like, I don't need that.

[Cara]

Right. And so like, if you're using it for. So I mean, this is not something I have personal experience with, but like a lot of people using it for their coding. Right, because, because it just makes you 10 times more efficient or 100 times more efficient. But like, does that kind of. This may even be a dumb question, but is that your code is amazing even if it doesn't happen even there? Or is that.

[Praksha]

Well, actually I had a little bit of experience with this. We did an AI coding week at Cargurus. What we did was as an engineering team, we paused work for a week and we all started working with these three AI coding assistants. Cursor, Windsurf and GitHub. Copilot.

[Cara]

Nice.

[Praksha]

Even people at my level. I was writing code with natural language during that week.

[Cara]

Yep.

[Praksha]

And as I was doing it, I was building a predictive model that would help me predict who was going to win the Masters. Oh. And as I was going into it, I was doing it completely. That's right.

[Cara]

You like golf. He likes golf.

[Praksha]

It was, it was, it was completely prompt driven development. It was writing all Python code for me and I was just prompting it and I was saying, oh, but what about this and what about that? And it would come back and say, oh, that's a great point. Let me incorporate that.

[Cara]

Oh, good. So you, because you have this skill and the knowledge you could all of a sudden become like hyper efficient because you knew, but you knew how to talk to it.

[Praksha]

That's right. And I've been trying to convince ChatGPT, get chat GPT to tell me that it's Skynet for the past, you know, three years. So I'm very efficient at prompting.

[Cara]

Oh, and how, how does it answer you on that?

[Praksha]

It hasn't yet told me that. It's, it's, it's judgment day. So I think we're in good shape right now.

[Cara]

I, I had a, a long conversation early on when I was using ChatGPT to create imagery.

[Praksha]

Uh huh.

[Cara]

I got into this philosophical conversation with it if it was an artist.

[Praksha]

Oh, interesting.

[Cara]

Yeah, yeah. And it would not say that it was an artist. It declined to define itself that way, which I thought was interesting.

[Praksha]

Yeah, that's really interesting. It's really cool to see how it impacts things like art because my sister in law, she's currently at MassArt.

[Cara]

Oh, very cool.

[Praksha]

And she's going there and we've had conversations about AI and like how they should be using it versus how they shouldn't be using it. Using it. And these are all conversations that every student needs to be having.

[Cara]

And it's, you know, in tech, technology supporting art is not new. Right? Like no, like you look at. And I won't have good examples in terms of the names of the, the artists. But like during the Renaissance, like there were certain. Probably was like, you know, Leonardo da Vinci or people who created like certain ways to do shadows or mirrors or lighting where it was like tracing and things like that. So even though when you see the end result, it's this most amazing thing ever, there was a little technology.

[Praksha]

Are you helping them graphic design today?

[Cara]

Right.

[Praksha]

It's just this techno, it's, it's a, it's a technology way to do art.

[Cara]

Right? Yeah, it's true. And it, and again, back to the democratizing. It democratizes art. Like I, I'll never be a musician, I'll never be a great artist, but it's fun to play with. So it gives you that creative zoom. Maybe not to be a professional, but to like play. Yeah, like have your voice, have your voice equalize. You can sing.

[Praksha]

Honestly, I'm going to stay away from that. I'm tone deaf. I don't even think AI can fix me.

[Cara]

Okay. You're beyond hope of anything there, but that's okay. Well, so what's next for you and for Cargurus? Are you working on any new cool stuff? Are you just perfecting your natural language search.

[Praksha]

We're just going to continue to forge ahead and take advantage of AI where we can and see kind of where this all goes, because this is a rapidly evolving space, and it changes on a daily basis, on an hourly basis.

[Cara]

Sometimes, I mean, it feels like sometimes you wake up the next you go to bed, and I bet as the head of engineering data, you deal with this all the time. You go to bed at night, you say, great, I have a plan. I know what I'm doing, and tell my team in the morning. Xyz, you wake up in the morning, you watch the news or hear the news or whatever, swipe the news, and it's all out the window. You got to start over.

[Praksha]

I have to. I have a daily TLDR AI newsletter that I read, that I spend time reading every single day from 5am to 8am I'm reading articles about what's happened in AI on the day before. Before.

[Cara]

Isn't that amazing? And the turn is so. It's just so much faster, right?

[Praksha]

It is.

[Cara]

And. And how does that. I mean, one last question before we go. Like, for your teams, right? So you have a lot of people who work for you, and you need to keep them feeling, like, comfortable in this space and all that. Like, how do you handle that as a manager of people who have to keep up with us all the time? Like, is it. Is it stressful? Like, what's it?

[Praksha]

I don't. I don't. I don't really expect them to have to keep up with it as much as I do, so I'm keeping up with it. And then when there are breakthroughs that I think we can take advantage of, that's when we'll start to train people up on those things. So we. My personal goal is to shelter them from some of that and be the person who has to put the extra effort in and then find the value and then help us, you know, see and realize that value as a team.

[Cara]

That's great. So you basically are relying on your human qualities of communication.

[Praksha]

That's right.

[Cara]

Filtering and watching out for your teammates to get you through this crazy tech time.

[Praksha]

That's it.

[Cara]

Cool. Well, with that, I think we'll wrap it up. And it's so good to see you. I'm so glad we met at John's event, and I hope you'll come back and tell us all the other cool things you're building. And I want to Hear more about WPI and Holy Cross for sure.

[Praksha]

Absolutely. 100%.

[Cara]

Okay. Thank you so much.

[Praksha]

Thank you. Cara, it was a pleasure.

[Podcast Host]

Thank you for joining us on Building AI Boston. Stay tuned for more enlightening episodes that put you at the forefront of the conversations shaping our future.