Audio & Transcript

Welcome to the Superconnector podcast. I am Matt Joseph, yourhost. We've got another round of startups to take a look at today from YCombinator's Winter 24 batch. Some very interesting startups doing some veryinteresting things. I'm excited to jump into it.

So first up is Scritch

empowering, independent veterinarians. Let's click through thisand see what we have here.

Okay. We empower veterinarians to start, run and grow their ownpractices. This is interesting because I've seen this done for a number ofdifferent health care providers. So think doctors of different types, you canimagine a primary care provider or even a specialist where they're part of ahospital framework, but they would like to be independent.

They want to control their own business. So I like this as abaseline.

So we've got a couple of founders. Claire and Rachel, sistersand co-founders of Scritch. We connect practices with new patients and providean AI powered platform to streamline operations, including scheduling, billing,and clinical workflows. We're excited to build a national network ofexceptional veterinary practices, starting with our first partner in LosAngeles county, California. I'm a big fan of LA. And we've got Lily Chen here.She's not one of the founders, but it looks like she is part of the earlypartner network that they're building. I think that's great that they've gotthat traction.

So let's jump into the problem here. Veterinarians and petparents both want a better healthcare system. Over the past decade, the influxof private equity and corporate acquisitions has dramatically altered thelandscape of veterinary medicine. And in the majority of cases thistransformation has contributed to a system that prioritizes volume over valueleading to dissatisfaction among not only pet parents and pets but alsoveterinary professionals, many of whom are leaving the industry altogether.

Yes. So this is actually a trend across all of healthcare whereyou have private equity coming in, buying and consolidating practices of aspecific specialty, and then selling the combined entity to a very large poolof capital or to a larger provider group. And this often leads todissatisfaction among the providers. In the case of doctors, it's the exactsame sort of question.

We see this happening across the entire framework ofhealthcare. Now we're seeing it happening with vets. I think that the problemhere is actually a really good one.

So they have some specifics on the problem. Veterinariansstarting and operating a veterinary practice in an industry with staffshortages, corporate competition and clunky software is incredibly difficult.And for pet parents over 70% of American households have pets and more peoplewant to invest in their pets, health and longevity. Many pet parents want theperks of a modernized high touch care model from local non-corporate clinics.

So I'm not certain that pet parents are sitting around sayingthere's too much private equity in the veterinary industry, but I do think thatveterinarians are definitely saying that. I agree completely with thatthinking. It is a big problem that you have corporate interests that arecontrolling how these providers operate. The reason that this is such a bigproblem, has to do with the fact that when these private equity interests comein, they essentially are trying to wring every last work unit out of theproviders, in this case the veterinarians, at the lowest possible cost.

There's very little upside. So the vets essentially go onsalary and they're now reporting to a private equity professional who knowsvery little about the practice of veterinary medicine.

Let's jump down to the team and the story. Claire and Rachelgrew up in SoCal and split up for college where Claire studied neuroscience andmachine learning at Princeton and Rachel studied computer science at Cornell.Before teaming up Claire trained models behind the full self-driving system inTesla's autopilot team as a machine learning engineer.

As a software engineer at Amazon, Rachel worked on one of theworld's most visited pages, the Amazon detail page, to provide the best buyingexperience for customers the globe. Scritch was born out of a personal desirefor highly comprehensive, stress-free veterinary care while navigating care forour 19 year old and 14 year old dogs. We learned of the immense challenges thatveterinary practice owners face. Now we're on a mission to superchargeindependent practice with world-class technology and resources and bring backthe warmth in veterinary medicine.

So you've got some great pedigree here. These are someinstitutions which ironically, I have some personal ties to. Princeton, which Iwent to as an undergrad and Amazon, which I worked at for a couple of years. SoI liked the pedigree. I liked the background. I liked the problem. They seem tobe outsiders coming into this industry, which I think that can work. I hopethat they have an incredible advisory board of veterinarians who they'reworking with. Some friends who are doing it that will help them get through thedifficult times. I liked the business. Claire and Rachel, good luck.

Alright, next up we have ego, an AI native 3d simulationengine. Getting us closer to the Matrix one pixel at a time. Well, now thesepeople are speaking my language. Getting us closer to the Matrix? We got tojump into this and see what this is about.

Yeah, let's jump to the TLDR. ego is an AI native simulationengine and platform where users can create and share 3d animated characters,worlds games, and scripted interactions, all using natural language. Imaginebeing able to prompt. Minecraft the Sims or Animal Crossing into existence. Andwe've got a little demo video here, so let's click through on this and see whatwe got.

Hi, I'm Shrek. Hold on. you're not scared of me. That's awelcome change. Wait a minute. You're not another princess who needs rescuing.Are you. And Please tell me you don't have a donkey sidekick.

No worries. Shrek. I'm definitely not a princess.

who needs rescuing. I'm Sailor Moon, and I fight my ownbattles. Now where are

Okay, let me pause this for a second. There's all kinds ofintellectual property issues on the surface here, because if they actually wantto use Shrek and Sailor Moon, they're going to have to license thosecharacters. This is obviously a demo. Maybe they already have the licenses inplace and I'm not aware of.

But we're talking about very likely games, which usecharacters, which the players have never seen. I think that's okay. I don'tthink people are necessarily wedded to characters. I think it's a cool concept.But it would be cooler if they could actually use characters like Shrek andSailor Moon.

All right, so let's scroll down here. The advent of largelanguage models and diffusion models presents an incredible opportunity tosolve three seemingly intractable problems in open-world games and simulations.Number one, it's hard for humans to write code and script interactions betweenassets characters and the environment to design fun games and experiences.Number two, it was impossible for NPCs to display human-like behavior and havetrue agency in these 3d worlds, NPCs being non-player characters. Number three,it's hard to create new 3d assets, worlds, and non playable characters.

Those are all legitimate problems.

I mean, those aren't necessarily problems that players have. Inthe sense that I think players are really just showing up to play something funand to have a game, but it's a huge problem for game developers. And if they'vefound a way to figure that out, that's great.

It looks like they have a few more demos here, which, we'regoing to skip over those and get to the point.

Why is this a big deal?

Since time immemorial humans have built tools to bring theirinternal worlds to life and the sandbox, the doll house, a tree branch shapedlike a rifle, all examples of tools we use to bolster our imaginative capacityto dream new worlds, characters, scenarios, and stories into existence. Theclosest we've come as a species to actualizing our imagination has been throughthe use of game engines to create simulated open-world experiences in games. AIis about to democratize the strongest tool we've had to realize ourimaginations and ego will be the simulation platform and engine that will powerthis new possibility.

I think there's probably a simpler way for them to describe whythis is interesting and cool.

This feels like they're talking to very specifically a marketof venture capitalists, which I think is the audience here. So, there's nothingwrong with that, but I do think there's a way to use more approachable languageto say why this is such an interesting and big deal.

All right. The team. Peggy and Vish are ex-Meta AI founders whoare huge game nerds. And we've got this great picture of them, just cheesing. Ilove that picture. They're having a good time. Before ego Peggy shipped thereal-time tracking pipeline for Meta's AR avatars and Meta Reality Labs, andpreviously researched autonomous driving behavior planning algorithms atStanford and Lyft. Vish led the Facebook AI Applied Research team and the MetaHorizon scripting team. He learned how to program by modding video games as ateenager, and also wrote screenplays for films.

So the team here is Ithink one of the big strengths. You have two founders who worked at Meta intheir gaming division building products that would give them deep insights inhow to actually do this, how to actually bring this to life. I think there's anopen question about whether they can actually create compelling games withthis.

I think that's where the rubber meets the road. But it's agreat team. It's an early stage startup so I don't expect them to figure it outfully yet. I like what I've seen. Congratulations to the ego team forlaunching. And I love the Matrix reference.

All right, let's move on.

Preloop. Deploy your machine learning models in hours insteadof weeks. Let's jump into this.

Preloop automatically translates your experimental scripts intoproduction machine learning workflows. The TLDR on this is Preloopautomatically translates your machine learning scripts into productionservices, handling the creation of the training pipeline and rest end points.This means science teams can focus on developing new models while cuttingdeployment times from weeks to just a few hours or less.

We're building Vercel for model deployments.

I like this. This is a huge problem with all the AIapplications that are coming out now. I think on the surface, it's prettyinteresting. Let's see what the founders are like. We're Tejas and Nikith andwe're building Preloop. Tejas previously worked as a data scientist andsoftware engineer at several companies, including Amazon. And most recentlyevolution IQ often leading zero to one projects on newly established teams.

Nikith has experience as a software engineer buildingmulti-tenant distributed systems. Most recently working at AWS on thenetworking team. So we've got another Amazon team here, which I love. Amazon isa great company. Produces exceptional engineering talent. And we've got a nicephoto of them hanging out outside of the Y Combinator office in mountain view.So, like that too.

Our goal is to unshackle scientists from the repetitive tasks,that accompany model deployments and empower science teams to move faster. Yes.Data scientists hate the work associated with deploying their models. Scienceteams spend anywhere from a couple of weeks to over two months deploying theirmodels. Assuming a team deploys 10 models a year and each model takes two weekson average to deploy. This is 20 weeks of science time spent just ondeployments. The biggest bottlenecks to quick deployment are lack of easy touse tools and delays caused by handing off deployments to a separate team.

This is very much aproduct which is built around having a large number of people who are workingon it. So we're talking about probably mid-sized to larger companies, maybesome larger startups. But probably not just two founders in a room working onand developing this.

We scan through your existing script, identifying keyinformation about the data transformations and the model being trained. This isused to construct both the training pipelines and data pipelines as well asinference end points that serve your model.

So basically we've got a couple of data scientists who aretrying to make it faster to deploy models. I think that's a great problem. Whatthey're asking for is intros to data science and machine learning managers atmidsize to large sized startups, as well as large enterprises.

So it's spot on what they're doing is going after largerorganizations that are trying to deploy models, they're using concepts thatthey learned while building this themselves. Good team. This is an executionplay.

Can they build fast enough? Can they get these largeorganizations to actually start paying them? They can. That's great business.

Congratulations. Let's move on.

Openmart. Get highly targeted local business leads using AI.

Okay. An AI alternative to ZoomInfo. Okay, so the TLDR, Heyeveryone! We're Richard and Catherine and we're here to present Openmart, aplatform that helps startups who sell to local businesses discover leads easilybased on their specific needs. Over the past few years. We have worked as brandoperators, the first product manager and the first engineer at high growth,SaaS startups. We crossed paths at Pinterest and have been good friends eversince. And we've got another nice photo outside of the Y Combinator logo.

Based on our conversations with sales teams from variousstartups, we identified the following challenges. Number one, finding SMB leadsis difficult because there's no way to locate them. People often have to searchrandomly through platforms like Google Maps, Yelp and Yellow Pages. Companieslike ZoomInfo charge upwards of $50,000 a year for an account. Filtering outthese SMB leads isn't easy.

Okay. So. This is something that I have some personalexperience with. I have bought leads both from YC companies that do that andfrom third-party providers to try to get the contact information for owners tostart a sales process. I assume that's exactly what they're doing here. They'relooking for ways to help salespeople automate the outreach process so that theycan spend more time selling and less time cleaning up leads. Lead cleaning is ahorribly intensive process to have to manage. Especially when you're a smallstartup. It's very expensive in large companies as well, although typically youhave specialists whose job it is to do that specifically. But the price point,if they can bring that price point down, I think they'll have a market. There'sa lot of people who want this. It looks like they have a little demo here.

Yeah, I like it. I think it's a smart business. The proof is inthe pudding on this one. I mean, can you actually generate high quality leads?If you can. It's great. Seems like they're technical. They have some goodexperience. Congratulations on launching Openmart and good luck to you.

Next up AgentHub. Automate any workflow with AI.

Okay. So, what do we have here? TLDR use a drag and dropworkflow builder to automate complex workflows with AI. They have integrationswith GitHub, Gmail, Outlook, Twitter, et cetera.

Try it for free using 45 plus ready to run automations or buildyour own automations for software development, sales slash CRM media slashnews, HR and hiring administration, financial analysis and more.

Okay. So that to me feels like they are boiling the ocean. Theybuilt a cool piece of technology, but they don't yet know who their customeris.

And so they are throwing it out there. I don't think that's aproblem. I think it's good to take a broad scope as you build your product totry to find exactly who the right user is. Although it puts you in danger ofbeing a solution in search of a problem. If you don't have a clear customer inmind, you have no customer. I think that's one of the big lessons in startups.And so I'm curious to learn exactly who they are going after in that mix of allthose different industries. No Code is a huge huge trend. That's something thatjust really picked up over the last few years. YC's had a few big winnerscompanies like Webflow which have done really well with no-code.

Zapier is another classic no-code tool. So there's a legacy ofthis at YC.

Being able to drag and drop. I think that's cool. That's a niceinterface. You just draw your connections. So this works like a lot of otherno-code systems. They have a few that they've put out there just as inspirationthey say. Github PR description Inserter. Multi-Article Summarizer. CandidateRanking System. Daily Stock Report Generation. I think they're probably goingto have to drill down on exactly what use case they want this to serve.

I don't think that people who are the kinds of early adoptersthey want are just sitting around looking for this. So this is an outboundsale. This is a manual sale. It's great they have great technology. They'regoing to have to build their sales motion out as well.

But I like it. I think this is something that's done well at YCand these guys seem to have some good technical chops. I don't see. Much aboutthe team here.

But. I do like the concept.

So congratulations to Max and Rahul on launching and good luckto AgentHub.

Next up. Omni AI. Structure your unstructured data.


Hi everyone. We're Anna and Tyler from Omni AI, and we helpclean up data using AI. Okay. So very similar to a previous AI startup, welooked at.

Omni inaction turn YCDemo Day videos into data.

Okay. So they take all the public Demo Day videos and extractsome of the most interesting business metrics, without needing to watchhundreds of videos.

Okay. That's pretty cool.

Okay. So let's talk about the problem. A company's mostvaluable dataset is often it's least accessible. Only 20% of corporate datalives in SQL with the remaining 80% locked away in unstructured formats.Reviews, chat logs, transcripts, PDFs. It takes some serious engineering workto answer questions like

graph the most common user complaints over the last four years.

What percentage of users complain about pricing on sales calls?

How many users are getting their questions answered by ourchatbot?

And this problem will only get worse as AI interfaces becomemore ubiquitous and more data is locked up in chat logs. Well, I like that. Ithink that's actually a very big problem. The data analysis is a huge componentof really any good modern business, but technology businesses in particular.And having to have data scientists write SQL queries to figure some of thisstuff out is both time consuming and very expensive.

I don't know if the play here is to replace the work that someof the internal analytics teams do. But if it could do that. Great. Andcertainly the point about chat logs seems to be a salient one for where themarket is right now. I don't know that AI interfaces are all going to requireyou to have unstructured chat data as the output, but for where we are rightnow, that seems like a very compelling problem. And if they were right aboutthat in the long run, that companies start using AI interfaces and there's lotsof data in chat logs then this could start to be very, very interesting.

So they have some more details on how they do it, but let'stake a look at the team. Anna comes from building ETL tools and data enrichmenttools at Hightouch. While Tyler has a background in machine learning andhealthcare applications most recently at Fair Square. We've dealt witheverything from sinking CRMs to fax machine APIs. We're building Omni becausewe hate working with messy data and we can make this problem a little bitbetter for everyone.

So, this is a problem that these guys have run into themselvesand their prior work. If they had some examples of things that they've actuallybuilt. I think that would actually help their case here, but this is a quick YConline pitch, so they don't want to belabor you with that. And that is fine. Iactually liked this business a lot. I think there's a very specific use caseand a specific customer that could be very excited about this. So Omni AI,congrats on launching. And good luck.

All right. Next up Duckie. Your AI technical support assistant.And they've got a duck. You know what? This kind of branding, I actually

really like, I'm excited about this.

Quack. We are Valerie and Joel. And we are happy to introduceDuckie, an AI technical support assistant that helps teams resolve issues inminutes. Okay. So this actually reminds me of

Clippy, which every few months or a year or so pops back up andpeople go crazy with the memes on Clippy, but Clippy was a pretty funassistant, a fun idea.

I think they may have killed it off. That was the Microsoftassistant from way back, but now we've got Duckie, so maybe we no longer needClippy. And we've got a social media post celebrating their first contractsigned. Well done. Oh, and we've got a video. Great. Let's dive into this.

For our listeners, theyare basically showing you a demo of how Duckie works. It's a pretty cool andslick video. Very highly produced.

They source everything that they provide. So that's nicefeature. Reminds me a little bit of Perplexity in the way that they do that.

Let's you update documentation. Okay, so this is good. I likeit. It's pretty highly produced, but I think with where the market is today, Ilike having real production value for videos that you put out for promopurposes.

If you can afford it, if you have the time for it. Caveats, butI like that. Duckie connects to your team's knowledge sources like slack, JIRA,confluence, century, et cetera, to find relevant information and generate issueresolutions in minutes, while writing back to your documentation, to keep theknowledge base up to date.

That is actually really, really cool. As you resolve issuesthat come up using Duckie, Duckie will also improve your documentation. And thedocumentation is important because engineers and other people who use theproduct are going to be using that to figure out exactly how to work withDuckie.

So I think that's actually really, really valuable feature.

Instantly understand your issue, know how to resolve it? Yeah,I think this is cool. I like this. This has some cool features. It's the basisof a very interesting assistant product, which could become very robust.

And the branding is on point. Well done Duckie. It can be verydifficult to put yourself out there with a brand that's very playful, butthey've done it and I think it works. I think it's memorable.

Okay up next. Eris biotech, developing cancer drugs that targettumors without poisoning the rest of the body. Well, we've seen

a startup that's doing that in our last round. But let's diveinto this one. Maybe there's something a little bit different here. We developcancer drugs that only become active inside solid tumors. By delivering activedrug only to the tumor site, we localize the effects of the drug to the tumorand minimize systemic toxicity.

Curing cancer. I love it.

All right. So let's learn about the team. We are Evita andRachel co-founders of Eris Biotech. We met working at the same lab at BYU whenEvita was a PhD student and Rachel was an undergrad. We collaborated on a peerreviewed paper and conference presentations. And we've been friends sinceRachel went to Dartmouth for a PhD and later worked at a startup developingcell therapies.

Evita went for a post-doc and a career in drug development inbig pharma. It would be nice to know which big pharma company that she workedat, but neither here nor there.

So, the problem. Cancer therapies are terrible. It's a racebetween who kills you first, the therapeutic or the cancer. At Eris Biotech, wewant to help with that starting with mesothelioma. Mesothelioma is a type ofcancer that grows on the pleura, the thin layer surrounding the lungs, with afive-year survival rate of less than 13%. Current mesothelioma treatments, notonly fall short in effectiveness, but are also systemically toxic. And mostpatients diagnosed with mesothelioma will die within 14 months after diagnosis.

I think they picked a really good cancer to start on. And itmakes a lot of sense, that they're going to try to build an entire platformwith this basis. With biotech companies, the question always comes much, muchlater as to does this work.

Can it get through all the phases of testing to become a, a,drug that's in clinical production? But I think the basis here is really good.I liked the founding team. And they seem to be approaching it in a way that'sfairly novel.

Eris Biotech develops drugs that only activate under low oxygenenvironments.

That's hypoxic environments. These drugs are referred tohypoxia activated pro drugs or HAPS. Low oxygen or hypoxic environment is ahallmark of the majority of solid tumors, including mesothelioma and healthytissues are usually not hypoxic. By delivering active drugs only to the tumorsite, we can use more potent drugs without systemic toxicity in cancertherapeutics.

Okay. So basically what they're saying is that by having drugsthat only activate in the tumor site, they'll prevent side effects that end upkilling patients, or really harming patients health over time. I think that'spretty cool. They're using AI drug discovery tools. And AI design compoundsfrom a library of over 2.6 million compounds. Okay. So I like this.

Congratulations Evitaand Rachel on launching. This is cool. The great thing about biotech companiesis that in every phase they go through of clinical trials, even without showingnecessarily that their solution works, they're able to raise more money andgrow the company. Expand what they're doing.

So if they're able to do that cycle, then I suspect they'llhave a great outcome.

Next up. Navier AI. Real time, CFD simulations.

Okay. TLDR Navier AI's machine learning, accelerated CFD solvermakes fluid dynamics simple and easy. We're supplementing traditional numericalmethods with machine learning models and a modern UX to enable engineers toiterate faster and build better products.

Navier AI was founded by Cameron and Evan, former SpaceX,rocket scientists with experience building software, hardware, and AI at theworld's leading companies.

Okay. So CFD is computational fluid dynamics. It's essentiallyusing technology to predict how fluid will move. That's broadly what's going onhere. I'm really interested to see where they take this because I haven'tlooked at this since I was in college.

Problem engineering simulation software is broken. Complexityoverload. Simulations are too slow. Repeating mistakes.

Okay. I could see that I haven't used any of these tools, but Ibelieve them when they say that. Navier AI is making CFD a thousand timesfaster with an ML based solver so engineers can simulate fluid dynamics in realtime. We train our model on high quality simulation, experimental data,providing engineers the reliability needed to design and optimize the missioncritical parts faster.

We're making fluid dynamics fast, easy, and accessible so thatany engineer can iterate, optimize without being throttled by their software.So I actually really liked this. They've got a nice little graph simulationhere showing I think exactly how this would work. And let's learn about theteam.

Cameron and Evan are aerospace engineers first joining forcesnearly a decade ago to start a liquid rocket propulsion organization at UC SanDiego. Together they fostered innovation across hardware and software atleading companies, such as SpaceX, Tesla, Aurora, General Atomics AnzuPartners, Astranis, boundary layer technologies.

We've got a nice photo of them hanging out in front of awhiteboard where their computers having fun on their Macbooks. They seem like ateam that likes each other.

They've worked together for a long time. They're working onsome really interesting stuff. This is frankly, a very niche business, but onethat I imagine is extremely lucrative.

I would want to look into the market for who the customersactually are for this. But they are rocket scientists, so I assume that they'regoing to figure out the sales motion once they get the technology locked in andthey seem to be building for themselves, which I think is great.Congratulations.

Navier AI on launching. Best of luck to you.

Next up. Veles. Your copilot for enterprise sales.

Unlock enterprise deals effortlessly with Veles. Veles is yourco-pilot for enterprise sales. Equip your sales team with an AI powered quotingtool integrated seamlessly with your CRM, CPQ and BI platforms. Give reps atool they will actually love and leadership the actionable insights they needto move the needle. Okay. Pricing enterprise sales is a headache. Fragmentedprocesses and tools result in subpar terms and lost opportunities. Many teamsresort to complex Google sheets, Slides / PowerPoints, manual CRM entries andemail chains just to handle pricing.

So pricing is very, very difficult and I think it's an areathat AI can probably make a big difference. It sounds like what they want to dois start with pricing as their moat. As their unique value prop and then find away to make this co-pilot work for other actions in sales. That would be myguess as to what the model is here because pricing alone may not be enough tobuild the scale of company that would justify them existing.

It might be. But it might not be.

It looks like we have a little bit of a demo here, so we'regoing to click through on this and see what happens.

Hello. I'm Simon with Veles. Veles is a co pilot for enterprisedeals that helps your sales team manage pricing and build more dynamic quotesthat lead to larger close run contracts. When your sales team logs into Veles,they'll be able to pull in all of their open opportunities from your CRM andusing your existing pricing model, build quotes on the fly. In this exampleI've created three options for my prospect to consider.

Okay. So. What they're doing essentially is helping salespeoplecome up with pricing when deals are complicated, which in the case ofenterprise deals they are. So you can imagine if you're a salesperson, you'reselling a piece of enterprise software that has a whole lot of bells andwhistles. You're about to charge these guys a million dollars for it. Youprobably want to have something that's going to help you flesh out all thedetails and all the pieces of that, that doesn't require a whole bunch of backand forth. Being able to instantly generate quotes. Specific to a customer, Ithink is really, really valuable tool.

So I like what they're doing. I think this could be veryvaluable.

Congratulations Veles. Good luck.

Next up Kabilah. Simplifying patient handoff for inpatientnurses.

We help nurses improve patient outcomes. All right. So anotherhealthcare business, I like healthcare in general. I'm curious to see whatthey're doing here. There've been a lot of plays that nurses in the recenthistory of startups. Picture this. You are an inpatient bedside nurse, startingyour shift entrusted with the care of 5 patients due to commonplace staffingshortages. As you get rapid fire handoff from the previous nurse, you jot downcritical information for each patient on a paper report sheet. Throughout yourshift, you're continually juggling urgent patient care demands, reading andupdating your scribble notes, attempting to communicate with inundatedphysicians and charting slash navigating data-dense notes within the EHRsystem. By the end of your shift, it is impossible to communicate all of thepertinent information about each patient in a rapid fire handoff. Thisperpetual cycle leads to increased medical errors, increased patient length ofstay and redundant inter clinician messaging.

So, this is a problem, which is actually a fairly significantproblem. The nurses take up a huge amount of slack for physicians who areseeing. Perhaps an even greater number of patients over the same period of timethe nurses are, and you end up with a need for coordination.

These problems that they're pointing out medical errors.Patient length of stay. Redundant messaging. These are all very expensiveproblems. The thing that I am curious to learn about is how they interact withsome of the other systems that are targeting nurses. think hospitals tend tolook at nurses as replaceable, even though they've gotten incredibly expensive.So I think this could be a very interesting opportunity to make nurses moreefficient and ultimately save the facilities, a lot of money to take some ofthe expensive costs out of the system and not have nurses waste so much time.The medical errors one I think is a really important point.

So let's learn about the team. We are Sarah and Umar. Sarah isa Stanford, MS / BS in computer science focused on AI, a 2022 Mayfield fellowsecond hire at EngFlow. Previously founded nationwide nonprofit organizationAWARE. Umar is a recent cancer survivor, a Stanford BS in computer sciencefocus in AI. He has a master's in management science and engineering and is a2023 Mayfield fellow.

So we have young founders who probably don't have a lot ofexperience selling into healthcare. I think this is going to be a verydifficult sales cycle for them precisely for that reason. That the people whohave authority in these matters don't necessarily view nurses as the targetgroup for new software.

The problem they're going to have is selling this software intohospitals where they're being pitched by quite literally thousands of startupsthat want them to use their technology. Standing out in that mix is a very,very difficult. And I don't see this working outside of a hospital context.

It's hard for me to imagine this becoming a big business basedon smaller facilities. Ultimately they're going to have to sell hospitals.

Maybe with the right nurse advocates, they would be able tomove the needle on this. They're going to have to find some environments wherenurses are very powerful and get those nurses to be the ones who drive adoptionof this. I think it's a big problem. It's just a question of, do the hospitalscare enough to actually pay and implement this sort of technology?

But. Kabilah. I wish you well, congratulations on launching.

Okay. So that's another 11 startups in the books. I think theyare all very interesting in their own way.

Thanks again for listening. Thanks for watching. And we willtalk to you soon.

Watch next