Description Transcript
Join David Hsu, CEO of Retool, as he opens Retool Summit 2025 with a visionary keynote on the future of internal software development, recorded live at SFJAZZ Center in San Francisco on October 7, 2025. You’ll get a first-hand look at how AI-native development is transforming enterprise AppGen, what’s on Retool's roadmap for empowering developers and technical teams, and hear real customer stories from executives at Uber, Motive, and Komatsu Australia.
Featured speakers include:
David Hsu (Retool CEO)
Abhishek Gupta (Retool Head of Product)
Burger Ebersohn (Uber Global Head of Partner Engineering)
Hamish Woodrow (Motive Head of Strategic Analytics)
Eric Cheng (Komatsu Australia Enterprise Architect)
Read more 0:03 Please welcome to the stage David Hsu, Retool CEO. 0:07 (upbeat music) 0:19 - Good morning and welcome to the first ever Retool Summit. 0:26 (audience cheering) 0:28 It is incredible seeing all of you today at SFJazz. 0:33 Looking out, I see customers, partners, community members, 0:39 all of you who are building the software that 0:41 makes the world go round. 0:45 Welcome. 0:51 For the past few years, we've been hosting virtual events 0:56 to show you all what's new at Retool. 0:58 And we called these Dev Days, and they were great. 1:01 But this year demanded something more. 1:06 We wanted to bring the Retool community together 1:09 in person to focus on one thing, 1:13 how to help you all build the best possible internal software. 1:21 And so we've packed today's agenda with amazing content. 1:26 You'll hear about our product Roadmap. 1:28 You'll get hands-on with labs. 1:30 You'll discover how organizations are solving previously 1:34 intractable problems with Retool. 1:37 And throughout the day, you'll get to meet 1:40 other Retool users. 1:42 And at the end of it, you'll enjoy a unique performance 1:45 that blends music and AI. 1:50 In just a few minutes, I'll be announcing something 1:53 groundbreaking, something that I think will change how every single one of you build and 2:01 retool. 2:03 The energy in this room is electric. 2:06 We are so excited to have you all here. 2:09 Let's get started. 2:13 So when we started to retool seven years ago, it was out of a deep, personal frustration. 2:19 My co-founder and I are both engineers. 2:22 And we spent so much time building internal software. 2:27 It was agonizing. 2:29 We were building the same stuff over and over again. 2:34 Admin dashboards, approval queues, customer support tools, 2:39 all of these had the exact same building blocks. 2:44 We were constantly reinventing the wheel. 2:49 And as lazy engineers, we thought to ourselves, 2:51 there's got to be a better way. 2:54 And so we built Retool. 2:55 And you can see, that's our first landing page up there. 3:00 The pattern to us is so clear. 3:02 Every internal tool required authentication, authorization, 3:07 database connections, caching, UI components, auto-lux. 3:10 It's so much more. 3:12 Developers were spending so much of their time 3:17 reconstructing identical foundations, 3:20 the same work repeated everywhere with zero leverage. 3:26 We were inspired by AWS. 3:30 AWS transformed programming by abstracting away servers, 3:35 storage and networking, freeing companies 3:38 to focus on their actual differentiation. 3:42 And so we believed the application layer 3:46 needed that similar kind of abstraction. 3:49 And so we built RETOOL for ourselves. 3:53 But pretty quickly, we learned 3:56 the problem was bigger than we thought. 3:58 Because it turned out we were solving a huge problem. 4:01 Internal software runs the world. 4:04 Every package that is delivered, 4:07 every prescription that is filled, 4:10 every mortgage that gets approved, 4:12 All of that runs on internal software. 4:16 It is the backbone of our global economy. 4:21 And yet, so many internal tools are slow, 4:25 buggy, frustrating to use. 4:28 It really is shocking how low the quality 4:32 of the average internal tool is. 4:37 Here, I've clipped a bunch of real news headlines, 4:41 showing just how bad it can be. 4:44 Check these out. 4:50 When the FAA system goes down, planes can't fly. 4:59 When your pharmacy, when the software doesn't work, 5:03 you can't get your medication. 5:07 Or when it's really bad, your doctor might tell you 5:10 then you have cancer, even when you don't have cancer. 5:15 What a shitty internal tool. 5:17 (audience laughing) 5:19 So as the Retro Business grew, 5:22 we learned that our platform 5:24 was not just helping developers build software faster, 5:28 but it was helping them build higher quality software too. 5:32 And so that is our goal, 5:34 to help you build the very best internal software 5:38 so you all can go run your businesses 5:41 and make the world go round. 5:44 That is what gets me out of bed every morning. 5:49 Let me just share two stories. 5:52 Paris, July 2024. 5:55 Imagine this, 10,000 athletes descending on your city, 6:02 206 nations, each with specific security protocols, 6:07 diplomatic requirements, religious dietary restrictions, medical equipment. 6:14 The math is staggering. 15,000 daily vehicle movements between three airports, 6:22 12 Olympic villages and 42 venues. 6:28 One delayed basketball team means we're scheduling an entire tournament bracket. 6:36 the traditional solution, an army of coordinators 6:40 with clipboards and walkie-talkies, 6:42 Excel sheets updated every hour, 6:46 printed but obsolete by the time they're printed, 6:50 WhatsApp groups exploding with messages 6:52 that no one can track, chaos. 6:57 And so the Paris Olympics in just a few weeks 7:01 created a living, breathing operations platform, 7:05 all in Retool. 7:08 Real-time flight tracking with GPS, instant delegation updates, 7:12 real-time calculations of ETAs, everything connected, 7:16 everything visible, everything instant. 7:19 This would have been impossible in a spreadsheet 7:23 or any other platform. 7:26 They had Retool apps for tracking every single flight, 7:28 every single bus, every single athlete. 7:32 And the result, the Olympics went off 7:35 a hitch. Everybody was in the right place at the right time. Even the French were surprised. 7:44 Retool didn't just move data, it moved athletes and helped deliver one of the world's largest events. 7:51 These are the kinds of internal tools that run the world. 7:57 Or consider the University of Texas Medical Branch. They have a big hospital system. 8:04 Prior to using RETULE, their pathology team could only process 50 toxicology reports a week. 8:13 That's it! So what happens to more than 50 people? Wanted reports? I don't know. I guess they 8:20 just had to wait, I guess. Well, their chief AI officer, Dr. Peter McCaffrey, decided to use RETULE 8:28 to solve this problem. 8:30 Without coding knowledge, he built a HIPAA-compliant, 8:34 self-hosted, Retool app that uses AI to interpret results. 8:39 Now, instead of delivering just 50 results per week, 8:44 they deliver 500. 8:46 That's a 10 times increase in patient capacity. 8:48 It means that more patients are getting the data they need 8:52 to make the right decision. 8:56 and not only are they delivering more results, 8:59 they're also more accurate than before. 9:01 This is software with real world human impact. 9:07 And that is what is so motivating to all of us at Retool. 9:14 And today, over 10,000 companies globally use Retool 9:19 to run their businesses and accomplish their missions. 9:25 And we are so grateful for every single one of them 9:28 and every single one of you 9:30 that's come to our summit today. 9:34 But as many of you know, software engineering 9:37 is undergoing a fundamental shift. 9:40 And technology has emerged 9:42 that it's changing how we all built. 9:45 Of course, I'm talking about large language models. 9:49 LMs are really smart. 9:51 They can be so helpful for software development. 9:54 They're so good at translating natural language into code. 9:58 So good at exploring a space of ideas 10:01 and generating prototypes. 10:04 So good at auto-completing code. 10:06 LLMs really speed up development. 10:10 And now, LLMs are being used to generate full applications. 10:14 Describe the app you want, 10:16 and in just a few minutes, you've got an app. 10:19 It almost feels like magic. 10:21 That's why we're seeing tons of new app trends startups 10:24 allowing almost anyone to go create a front-end app. 10:29 These tools are amazing for inspiration, 10:32 but that's the thing. 10:33 In most cases, they only create prototypes. 10:38 That's because if you look under the hood, 10:41 you'll see they're missing what's most important to enterprises. 10:44 They don't connect to your real data. 10:46 They don't understand your enterprise security policies. 10:50 They don't use your design systems. 10:52 They just aren't built for the enterprise. 10:56 And so what happens if AI is deployed in your business 10:58 without the proper guardrails? 11:01 It can create some pretty serious problems. 11:04 For example, it might go rogue and just 11:08 drop your production database. 11:11 It's so bad that we're kind of embarrassed showing 11:13 the name of the company. 11:14 So we had to blur it out, actually. 11:16 But it's a vibe coding product you've probably heard of. 11:20 And so while LLMs are great for some things, 11:23 they're not so great at others. 11:25 Generative AI is fast and creative, 11:27 but if it's not connected to your business data, 11:30 it's just not really that useful. 11:32 And if it doesn't obey your security guardrails, 11:36 you just can't use it. 11:39 These AppGen products are really great 11:41 for building a fun app for your kids, 11:43 for your friends, for prototyping, 11:46 but that's not enough. 11:50 Because what all of us care about in this room, 11:52 what all of you care about is production internal software. 11:58 And while the bar to prototype has dropped, 12:02 the bar to shipping a production application hasn't changed. 12:06 What enterprises really need is a trusted, scalable, 12:10 secure platform to build with AI inside of their organizations. 12:16 Well, I've got great news, 12:18 Because that's exactly what Retool 12:20 has spent the last six years building. 12:23 We've been obsessed with solving boring, 12:25 but really hard, last mile problems with software development. 12:30 The data connectivity, security, governance, deployment, 12:33 the very things that turn an LLM generated application 12:38 into a production ready application. 12:42 Think of it this way. 12:44 What generative AI is bad at is precisely what 12:48 Retool is good at. 12:50 This makes Retool the perfect complement for generative AI. 12:56 That's why I'm thrilled to make a big announcement today. 13:01 Retool is now the world's first enterprise AppGen platform. 13:08 [APPLAUSE] 13:12 [APPLAUSE] 13:15 Just a few minutes ago, our engineering team 13:17 released a new capability that allows you to generate 13:20 applications in Retool just by typing a simple prompt. 13:24 If you visit our home page on your desktop computer, 13:28 you'll see the newly updated version. 13:30 And you can quickly type a simple prompt, let's say. 13:34 Create an app that helps you manage chore assignments 13:37 for my three kids at home. 13:39 That problem goes directly into the RUTL product. 13:41 And in just a few minutes, there it is, just a few words. 13:47 We've created a working application. 13:50 A parent can click around, change shore assignments, 13:53 add new chores, mark them as completed. 13:56 It even created a new database to store changes. 14:00 Pretty cool, huh? 14:01 What do you think? 14:04 [APPLAUSE] 14:05 (audience applauding) 14:09 Well, no, not really. 14:13 This is not very cool. 14:15 This is old news, actually. 14:18 Who cares about building this fun application 14:22 for tracking chores? 14:23 What would make this really cool 14:25 is if we could build a real production application 14:29 that actually runs your business, 14:31 that actually integrates with your production data, 14:35 that actually is security built in that is hosted in your cloud. 14:40 Not a cloud environment you don't even use. 14:42 Wouldn't that be cool? 14:44 Well, let's try it. 14:46 Let's go back to the home page. 14:48 Let's say that I work in a hospital, 14:51 and I want to go build an application. 14:53 Build me an approval queue for doctors 14:55 to approve prescriptions using my Postgres data. 15:00 There it is. 15:01 This isn't just a front-end app. 15:03 It actually writes back to your database. 15:06 Retool will search for your distinct resources. 15:09 It will find the relevant tables, 15:11 and it creates an app on top of that. 15:13 And when you hit the Approve button, 15:15 it actually saves to your database. 15:17 And the prescription status will change to approved. 15:20 What's more, this is a Retool app. 15:23 And so it inherits all the security policies 15:25 you've already set up in Retool. 15:28 So if you're in a hospital, 15:29 only doctors will be allowed to use this application. 15:32 that every time they press approve, 15:33 it is automatically logged in your audit logs. 15:38 We believe this is the future of building internal software. 15:44 This is what is available today in Retool. 15:47 And we think this is how internal tools will be built 15:49 at every enterprise. 15:54 That's because if you look under the hood, 15:56 you'll see there's three critical attributes 15:58 that no other platform has. 16:01 First, it's on your data. 16:03 So Retool connects directly to your production databases 16:06 and APIs, and we understand your actual schemas, 16:09 your customers, your orders, your prescriptions. 16:11 And so the generated apps are relevant from the very first 16:15 prompt. 16:16 Second, what in your cloud? 16:19 Whether you use our secure multi-tenant cloud 16:22 or self-host Retool, your data never leaves your environment. 16:25 You can generate apps in your cloud on your data. 16:30 And third, it's secure. 16:33 Because it's a Retool app, it automatically 16:35 inherits all your org security policies already set up in Retool. 16:40 It hooks in with Okta, your role-based access controls, 16:44 and logs everything to the audit logs. 16:47 You will never have to reveal secrets to people building 16:51 a Retool because you didn't have to. 16:53 It's all handled automatically. 16:55 Governance is not an afterthought. 16:56 It is the foundation. 17:00 Over the last few months, we've invited around 100 customers to play around with building, 17:05 with AI and Retool. 17:07 And Camilla, the chief operating officer at Brex, said this, "At Brex, speed and security 17:13 aren't trade-offs. 17:14 They're requirements. 17:16 By adding AI that understands our actual data and security requirements, our teams can move 17:24 It's start of velocity while maintaining bank grade governance. 17:28 That's the power of an enterprise AppGen platform. 17:32 These three attributes on your data, in your cloud, secured by default, that is what we 17:37 believe defines an enterprise AppGen platform. 17:42 But AI is transforming not just how we build, but also who builds. 17:50 We originally built RETool to help engineers. 17:53 That's because my co-founder and I were both engineers. 17:56 We couldn't find a product that combined the power of code with a composable building block. 18:02 That's something like a visual basic offered. 18:04 And so that's why we built Retool. 18:06 We built it and we invested a lot in power, flexibility, and escape hatches. 18:11 And that's why engineers love Retool. 18:14 And we know many of you here are software engineers. 18:21 But we also know that many of you are not. 18:26 In fact, we ran a survey earlier this year, and we're seeing a pretty fundamental shift 18:32 in who is building internal software. 18:36 And our survey, over half the builders were non-software engineers. 18:43 that your colleagues in data operations and product, and they are changing how we work. 18:55 And over time, we actually expect that even more non-developers will be building software. 19:01 This is critical because people and teams are being asked to do ever more with AI. 19:07 In fact, a study found that over two-thirds of companies had issued AI mandates. 19:15 Here's some really great news. 19:17 When builders have access to enterprise AppGen platform, they don't just meet expectations, 19:24 they shatter them. 19:26 Our survey revealed that 74% of retail builders report exceeding their leadership's productivity 19:34 expectations. 19:35 With Retool, problems that took months now take days. 19:42 I'm so inspired by one of our customers, Charlotte. 19:46 Charlotte was an early Retool builder. 19:49 And she's been promoted four times in four years 19:53 because she's a builder and she solves problems. 19:56 That is the power of Charlotte, plus an enterprise grade 20:01 and AppGen platform. 20:06 So, really excited to have you all here. 20:12 In the early '70s, when Apple just got started, a journalist asked Steve Jobs what the purpose 20:20 of a computer was, and if it was true that all of us one day would have a computer. 20:26 And Steve Jobs gave the analogy of, well, it's like a bicycle for the mind. 20:30 But what he said after was that Apple is fundamentally a tool-making company. 20:36 And Apple's mission is to create tools for all of their customers, such that all of their 20:44 customers can leave a dent in their corner of the universe. 20:50 And that's how I feel about Retool as well. 20:52 That really resonates with me. 20:56 mission is to help you create tools such that you all can accomplish your missions, make 21:06 the world go around and leave a dent in your corner of the universe. 21:13 Whether it's treating more patients or transporting athletes for the Olympics and delivering a 21:19 great global event. 21:23 in this room is a builder. 21:28 And you all are building the future. 21:32 Let's go build. 21:35 (audience applauding) 21:42 Thank you all for coming. 21:43 We have so much more to show you. 21:46 I'll now welcome to the stage Abhasek, 21:48 who will walk us through our exciting product roadmap. 21:51 Thank you. 21:52 (audience applauding) 21:56 Please welcome, Retool Chief Product Officer, 21:59 Abhishek Gupta. 22:05 (audience applauding) 22:10 - Thank you, David. 22:12 And thank you, everyone, for coming 22:13 to our first ever inaugural Retool Summit. 22:17 We're so excited you're all here today. 22:20 I'm Abhishek Gupta, Chief Product Officer here at Retool. 22:24 And I'm really excited to walk you 22:25 through where we're going today. 22:30 AI is transforming businesses. 22:32 It's creating an explosive demand for internal software. 22:37 But there's a catch. 22:40 If not harnessed correctly, it's promise 22:42 of greater productivity, more competitiveness, 22:45 and more innovation may never materialize. 22:49 Retool is a trusted platform that'll 22:52 help you build and deploy A applications. 22:56 We want to help you and your business win. 23:00 So I want to show you a little bit 23:01 about what we've been cooking this year, what's next 23:04 and what to expect. 23:08 Now, when you're in an enterprise, 23:10 this is typically the software development lifecycle 23:12 you follow. 23:14 First, you build, then you scale, then you manage, 23:19 Then you automate. 23:21 And for the last six years, we have been shipping capabilities 23:25 every single day to support this. 23:29 In fact, just the last year alone, 23:31 we released over 75 new capabilities across security, 23:35 governance, integrations, and one of my personal favorites, 23:39 Multi-Page, to help people with some really complex 23:41 applications. 23:44 And this cycle is what a lot of builders like you 23:47 do every single day. 23:49 And so our roadmap is built for you. 23:52 And so no matter where you are on this journey, 23:54 we're going to support you. 23:57 And this is how we're going to structure our conversation 24:00 today. 24:01 First, we'll talk about how do we help you build, 24:04 taking your ideas into apps instantly. 24:08 Two, we'll talk about how you scale consistently. 24:11 How do you make sure that everything you're building 24:13 is consistent and govern across your organizations. 24:17 Three, we'll talk about how do you manage confidently? 24:21 How do you make sure everything is secure 24:23 and enterprise ready? 24:24 And then finally, we'll talk about automating work. 24:27 How do you transform your businesses? 24:30 So at the end of the day, you'll have a really good idea 24:32 of where we're going. 24:35 Let's start with building instantly. 24:38 How do you instantly and quickly take your ideas 24:41 into production. 24:44 Let's start with a problem that I think all of you in this room 24:46 know far too well. 24:49 Your best ideas are stuck. 24:51 They're stuck waiting on someone else, 24:55 stuck on a spreadsheet, stuck on the shelf somewhere. 25:00 A business analyst I was speaking to more recently 25:03 at an e-commerce company told me that we had six months 25:05 to get internal dashboard built because of how constrained 25:08 their development resources were. 25:10 That isn't just inefficient. 25:11 It's a complete drag on your business. 25:14 Decisions take forever. 25:16 Competitors are moving faster. 25:19 And your best people, the ones closest to the problems, 25:22 are blocked. 25:23 And they might leave. 25:24 And everybody hates it. 25:26 Your best ideas should not have to wait for anyone. 25:32 At Retool, we've always focused on helping builders just build. 25:36 That's been our goal since day one. 25:41 And when you land here, this page that you're all probably too familiar with, we make it 25:46 super easy to build an app. 25:48 We make it easy to get started and make it easy to ship something to production. 25:54 But guess what? 25:56 It's about to get faster. 25:58 It's about to get better. 26:00 As David mentioned earlier, we are now an AppGen platform. 26:05 So now, when you land in Retool, you'll see Assist right here in the IDE. 26:12 And all you have to do is use natural language to kick things off. 26:18 It works, and it works for you, building an app instantly. 26:22 Let me show you how it works. 26:25 You can give it a prompt. 26:28 Let's say I want to build a customer 360 dashboard across my Salesforce, Postgres data. 26:35 So in this case, I gave it a prompt. 26:37 I wanted a customer dashboard, where 26:39 I want to search, add, modify records on top of my Postgres 26:43 database. 26:45 All you have to do is enter it, and then we get to work. 26:49 We start reasoning with your prompt. 26:51 You can see we start understanding and sharing 26:54 a plan for execution. 26:56 Not just that. 26:56 It understands your table schemas. 26:58 It scaffolds a UI. 27:00 And your queries are now automatically written. 27:03 And your components are coming together. 27:06 And so instantly, you have a production-ready application 27:09 showing you how to use and access information 27:14 about all of your customer data. 27:18 And now that you have an app, it's working. 27:21 So your team can use it. 27:23 You can add data to it. 27:25 You can-- it'll understand all of your logic 27:28 across your filters and dropdowns. 27:30 You can even start editing information, 27:32 which you'll see here. 27:35 Again, this is not a disposable prototype. 27:39 This is not a family short tracker app. 27:42 It's an application working on your data, 27:44 on your business logic. 27:48 But let's say you wanted to go a step further. 27:50 Let's say you wanted to brand it with your company's internal 27:54 branding. 27:55 Well, we've got Spotify in the audience today. 27:58 Before I'm done even talking, it already branded it 28:00 as a Spotify branding, which is amazing. 28:07 But let's say you know how to use Figma, your designer. 28:11 Pretty soon, all you have to do is just upload a Figma file. 28:15 And we will make your application look beautiful. 28:19 And you'll be ready to go. 28:26 And because Retool is the only platform 28:28 where you can combine the power of code, natural code, 28:32 drag and drop, and natural language, 28:35 you will be able to simply click on containers 28:39 and articulate and write what you want to do, 28:41 and it will apply for you automatically. 28:45 And if you want to collaborate with others, 28:47 well, your team can also leave comments right here, 28:50 and then we'll apply it for you. 28:52 All these things are done automatically. 28:56 This is the power of Retools AppGen platform. 29:00 And the first time these AppGen capabilities 29:04 can be used to bring software into production, 29:07 to deliver your software into production. 29:10 It's on your data, on your cloud, and it's secure. 29:15 And I'm so excited to share that this assist is now 29:18 available to all of you today, all of our cloud customers, 29:22 and all of you using us on Self-hosted. 29:27 Now, you heard David mention earlier that we had about 100 customers using our private 29:34 beta for this product earlier this year. 29:36 One of those customers is Motive. 29:38 Motive is an AI-powered integrated operations platform that drives meaningful and real-axiom 29:46 reductions with drivers on the road. 29:49 Hamish Woodrow, head of analytics at MOTIV, 29:52 leads a team that supports thousands of users. 29:56 But what's really great about Hamish, 29:58 he's a great example of this new class of builder. 30:01 He's a business strategist, but a chemical engineer 30:04 by training. 30:05 And he's now shipping internal production software, 30:07 which was unheard of years ago. 30:10 So please, allow me to welcome Hamish to the stage 30:13 to tell us more. 30:15 Please welcome Hamish Woodrow, MOTIV, 30:18 Head of Strategic Analytics. 30:22 [MUSIC PLAYING] 30:27 Thank you very much, Abhishik. 30:29 And so today, I'm going to cover a little bit 30:32 on my philosophy and overall approach 30:34 to how to actually integrate AI into data teams 30:37 and actually be effective to scale it up. 30:40 And the base, our ambition, is to build 30:43 world-class internal data products. 30:45 That's what we really want to go out and do. 30:48 And at MOTIV, we have a really interesting set of data 30:51 to do this with. 30:52 So we have hardware data. 30:54 We have vehicle telematics data. 30:56 We have usage data on applications, on mobile, on web. 31:01 We have our traditional CRM. 31:03 We have subscription data, huge wealth of data 31:05 that makes it truly fascinating to work with as a data 31:08 professional, but also makes it very complicated 31:11 to deliver to front-end teams really high-quality applications 31:17 or value. 31:18 And I think this is, in general, one of the problems 31:20 that I spend a lot of time thinking about 31:22 is how do you actually leverage AI in a scalable way that 31:27 actually delivers value to the business. 31:29 And this is something that we've been working on 31:32 and what we've been working also with the Retool Team on. 31:35 And one of the big bottlenecks we have 31:37 is the analyst in the loop. 31:39 Because of the complexity of the information 31:41 that we're bringing together, we always 31:43 have the analyst and data professional 31:46 in building that data model, really understanding 31:48 the business domain, the industry, 31:50 how to link all this data together. 31:53 And that's always a bottleneck towards getting and delivering 31:56 value to kind of our customers, internal customers. 31:59 And so this application layer has taken over a year 32:03 as many different forms from traditional software 32:07 to custom-built applications. 32:09 But actually, the delivery of this and the amount of people 32:11 to deliver it is very complicated. 32:14 And so this is kind of our approach. 32:17 What does it actually mean to build world class applications? 32:20 So really, these are the four components 32:22 that we think about, which is, at the base, functionality 32:26 and content. 32:27 That's everything that's like the completeness of data. 32:29 That's the consistency, latency of data. 32:32 You've got performance and stability and security, 32:34 foundational to any scalable products. 32:37 In reality, you can build a lot of products. 32:39 But if they don't have this middle layer of performance, 32:42 stability, and security, you never 32:43 scale it to thousands of people. 32:45 And that's always been an impediment 32:46 in getting good quality applications out there, 32:49 has been this layer. 32:51 And then you start going beyond this. 32:52 It's usability, user experience. 32:54 Like, if you want to build high-quality applications, 32:56 it's got to be interactive. 32:57 It's got to really think through the workflows 32:59 that people do on a daily basis. 33:01 How do you improve click-throughs 33:03 on all the application? 33:04 How do you get the right information to the person 33:06 in the right way? 33:07 And this old one was always leveraged-- 33:09 some sort of JavaScript or some sort of coding 33:11 that usually moved far beyond what a data professional is 33:14 going to be knowing deeply about. 33:16 And finally, design and aesthetics. 33:18 Truly, if you want a great application 33:20 to be used and adopted widely in the organization, 33:23 you have to be consistent in your design. 33:25 You have to be consistent in branding. 33:26 You have to care about it. 33:28 And that really is the difference between what 33:30 is functionally a good data product 33:34 to what is actually a used data product 33:36 and highly adopted in our organization. 33:38 And so when we look at it, ultimately, 33:41 The analyst strengths, the data professional strength, 33:43 is in functionality and content. 33:45 That's always their bread and butter. 33:48 But as you move up, their strengths go away. 33:51 And you won't find the value delivered. 33:54 And the reality of these things is the last mile problem. 33:56 The value generated is really that final application. 34:00 Yes, the data is the foundation to it. 34:02 But you can't deliver it if you're not 34:03 getting people to use it. 34:05 And so this is kind of how we've been thinking 34:08 about how to apply AI. 34:09 It's like we want the people with the greatest 34:12 knowledge of a problem, the data models, the insights, 34:16 building ML models on top of it to be 34:18 able to deliver products to our internal users 34:22 in a much more effective way. 34:23 And in general, this is how we want a data professionalist 34:26 act to really increase their impact. 34:29 And so what we see is performance and stability. 34:33 That's the kind of what we're looking for in terms 34:36 to this walled garden of development within Retool. 34:39 We want people to go out and build, 34:41 but we want it to be done in a secure way. 34:43 Or else what we run into is this problem of applications 34:47 never scaling, because they get blocked at security, 34:49 or you can't enforce role-based access control. 34:52 You can't enforce write permissions. 34:56 You want to make that usability and user experience come in, 34:59 and that's effectively using different coding. 35:01 And historically, you'll have a data analyst, data signers, 35:05 copying code from different code snippets, pasting stuff in, 35:09 usually makes a pretty clunky application. 35:11 Doesn't really deliver. 35:12 It's not there like prime value add. 35:14 And so again, this is something we 35:16 see AI being transformational and actually 35:18 extending that capability of someone who's 35:20 coming up with the ideas. 35:22 And finally, the design and aesthetics. 35:24 Universal design. 35:25 I think about it a lot in terms of how can you, 35:28 especially as you end up with very distributed teams 35:31 across numerous geographies, interacting 35:33 with many internal applications team, 35:35 how do you get consistent design and aesthetics? 35:38 And that's where AI is much better, 35:41 to be your oversight in terms of enforcing branding, 35:44 enforcing design principles that are 35:47 consistent in every product you see out there when it comes 35:51 to what you may sell to customers, 35:53 but is often not reflected in internal products. 35:55 And that's really important. 35:57 So all this is really extending the capability 35:59 of our internal data professionals. 36:01 And we're already using Assist Today 36:03 to exactly do that, to extend out the functionality, 36:07 modify existing applications, improve the overall UI, 36:11 and just also ask it optimize the application you have 36:14 to hand that's already been built. 36:16 And so that's where we're seeing a lot of value 36:18 and extending things out from our point of view. 36:21 And I think this is something that we think about 36:23 in the custom application world is, 36:25 is often we've always been limited 36:27 by having the functionality and the content, 36:30 but we're trying to put things out there in dashboards, 36:32 that we're trying to figure out the right way 36:33 to get information out there. 36:35 In the software world, you're generally 36:36 starting with this well-designed product at the top. 36:39 But often, you're limited by that bottleneck of the analysts 36:42 providing the integrations, the data, the SQL queries, 36:45 the information. 36:46 And that always breaks down over the long run. 36:48 And so what we see is by combining 36:50 AISIS with the knowledge of our data professionals, 36:54 we're really being able to actually deliver value 36:56 and actually scale up our products much more effectively. 36:59 So it's something I'm super excited to see 37:01 development of, this is actually one of the areas that I really see AI getting 37:05 actually real world impact in terms of the value we can deliver as a team and 37:10 really being great to work with the team here. And so with that I'll welcome 37:16 back on the stage I think Abhishek. 37:24 Thank you, Hamish. It's amazing to see all the great things that Motive has 37:33 accomplished on Retools App Gen Platform. We're so excited to continue 37:37 supporting them on their journey. Now, we mentioned earlier that Hamish is a 37:44 great example of this new class of builder, someone who's percolating to all 37:48 parts of an enterprise. 37:50 And we also know earlier that about 50% of our builders 37:55 are developers, software engineers. 37:57 In fact, we looked at the titles all of you attending today. 38:01 We saw about half of you are also software engineers. 38:05 And that's because you're builders. 38:08 But the other half of you are domain experts. 38:12 You go to market engineers, business analysts, data 38:15 analysts, operations experts. 38:19 And you're close to your business problems. 38:22 You're both motivated and able to solve these challenges 38:25 by building applications. 38:28 And Retool is a platform that's you be a builder too. 38:32 And that's because we just made it so easy for you 38:34 to ship applications. 38:36 We've shrunk that gap. 38:37 We're an idea to production. 38:41 Let's move on to the second piece. 38:44 How do you scale with consistency? 38:47 This is important because when building gets this fast and easy, 38:50 a new challenge emerges-- 38:53 chaos. 38:56 When you unleash builders across your organization, 38:59 you get more apps and more innovation. 39:02 But you also get duplicated logic, inconsistent UIs, 39:06 and teams just constantly reinventing the wheel. 39:09 And so it's just scaling innovation. 39:11 You start scaling complexity. 39:13 With Retool, we want to flip that equation. 39:16 With Retool, innovation should compound. 39:19 It should increase quality, not chaos. 39:23 So to help you scale Retool across your organization, 39:26 we're going to talk about three new capabilities today. 39:29 And we'll release these in a structured way that 39:32 all build on top of each other. 39:36 So the first one, reusable functions. 39:40 This is all about creating and sharing complex logic 39:44 so that you can finally eliminate 39:45 that duplication in organizations. 39:48 This is all about ensuring that when you build something once, 39:53 you can use it everywhere. 39:55 Because nobody likes to rewrite the same query 39:58 in every single new app that they create. 40:01 And to use it, we'll make it accessible for you 40:03 right here in the canvas. 40:06 You can start chaining all kinds of complex things here. 40:09 And you can think of these like queries, 40:11 but just much more powerful. 40:14 You'll be able to bring in multiple different data 40:18 sources together. 40:20 And because they're running on the back end, 40:22 they're going to be more performant and more secure. 40:24 And you can just do more complex things. 40:28 But importantly, these will be shareable across not just 40:31 your apps, but agents and workflows too. 40:37 Next, we'll launch libraries. 40:39 Libraries will allow you to store all your Retool 40:42 primitives as building blocks so that you can easily discover 40:46 and share these across your organization. 40:52 And to use these, it's as simple as just asking 40:55 Assist to import them. 40:58 And what's great about libraries is 41:00 they'll use everything that you've already built in Retool-- 41:03 components, modules, workflows, agents, functions. 41:08 You'll be able to version and permission your libraries 41:11 so only the right people and the right systems 41:13 can access them. 41:14 This is how we believe you'll enable consistency 41:17 across your organization. 41:21 But we go a step further. 41:24 What if your builders didn't have to think about queries 41:27 or tables at all? 41:30 Let's take a look at semantic objects. 41:34 Semantic objects will allow all the builders, like Hamish, 41:39 tomorrow's builders, to build secure applications on Retool. 41:44 And it will be by default. 41:45 How? 41:46 Because platform teams will be able to define 41:48 the actual objects, like employees, orders, customers. 41:54 And these builders, what they will get out of this 41:58 is a guarantee on security, correctness, and consistency. 42:03 So let me show you a little bit about how it works. 42:11 First, we'll help you stitch together all your data 42:15 into a single place. 42:16 So let's say, for example, you want 42:17 to build an employee's object. 42:20 You can bring your data together from HR systems 42:23 like Workday, recruiting systems like Greenhouse, 42:27 payroll systems like ADP. 42:30 We'll be connect all that data together. 42:35 Then we'll help you clearly define the logic 42:38 that allows you to take actions on top of this data. 42:41 So let's say, for example, you had an employee record that 42:44 was updated in one system or added. 42:46 We'll help you build a logic that allows you 42:49 to update another system. 42:53 And then finally, you'll be able to define permissions. 42:56 Permissions will define who and what can access data 42:59 and what they can do with it. 43:02 And so now, a human resources manager, 43:04 when they look at this app or they use this app, 43:07 they'll need access to employee payroll data. 43:09 But an IT manager does not. 43:12 And so each semantic object will bundle together 43:16 the data, the logic, and the permissions together. 43:20 And so now, making your builders stitch rock queries together, 43:24 They'll work with objects like these-- 43:26 employees, orders, customers. 43:29 And for the first time, that means an employee 43:31 needs the same thing to someone in HR, IT, or finance. 43:36 And a mixed retool feel more like a language 43:38 that your whole business can build with. 43:41 That is a core idea beyond semantic objects-- 43:45 reusable building blocks. 43:49 Now, we believe this is truly groundbreaking, 43:52 because you can now build apps at a much higher level. 43:55 And so to use it, it's really easy. 43:57 When you're building an app, you can just 43:59 point your table source to this employee data, 44:02 and everything is built in from the beginning. 44:05 And let's see what happens when you publish it for others to use. 44:10 When you view it as a head of HR, 44:13 because you have access to a lot more things, 44:15 you can see all kinds of sensitive data, 44:17 like compensation data, performance ratings 44:19 for every single employee. 44:22 But let's say you're a manager. 44:25 You can actually view far less information, far less sensitive 44:29 information. 44:31 And it depends on who you're managing 44:33 who you're not managing. 44:36 And then when an employee sees someone's profile, well, 44:38 they get access to far less information. 44:41 The beauty of this is that the builder who built this 44:44 didn't have to think about any of this. 44:45 It just worked. 44:48 Permissions followed the role automatically. 44:50 Security and governance were built in from the start. 44:53 And RAI followed the guardrails. 44:56 And so now, trust was built into every single interaction 44:59 with semantic objects. 45:03 And so what's really great is functions, libraries, 45:06 and semantic objects will all build on the foundations 45:09 you've already created in Retool, 45:10 and they'll build on each other. 45:12 That's how we believe you'll scale across your organizations. 45:16 Instead of chaos, you'll have order. 45:18 And everything you create in Retool 45:19 finally strengthened the whole ecosystem. 45:25 So we've covered how to build instantly, 45:29 how to scale consistently, but how 45:31 do you make sure that everything you build 45:33 is secure, reliable, and production ready? 45:40 We all know from you that the last mile of software development 45:44 is always the hardest. 45:46 That last 10% sucks. 45:49 Sometimes it takes just as long as the first 90%. 45:52 It's where all the pesky and long tail issues show up-- 45:55 bugs, performance issues, meeting your security standards, 45:58 deployment issues. 46:01 We just don't think builders should 46:03 have to solve for any of this. 46:05 And our philosophy here is very simple. 46:08 The platform should solve it for you. 46:12 And so to do this, we're going to release 46:15 three new capabilities to help you manage confidently. 46:18 First, expert agents to secure and optimize your apps. 46:24 Two, native testing to ensure quality at scale. 46:27 And then three, a brand new hosting option 46:30 for the ultimate flexibility. 46:33 Let's talk about the first one. 46:36 We're going to first introduce expert agents. 46:39 These agents won't be used for helping you build your apps. 46:42 You can think of these more as specialized agents. 46:45 Special Forces running 24/7 in the background, 46:48 constantly focused on making your apps better. 46:52 They're going to raise the quality bar on your apps 46:54 automatically. 46:56 Let's look at the first one. 46:59 You can ask an expert agent to just help you 47:00 when you're building an app. 47:02 This is our performance agent. 47:04 It found an optimization to make on a query. 47:07 And guess what? 47:08 It was pretty significant in improving the loading time. 47:11 And more importantly, it explained what it did. 47:14 So the next time, you won't repeat that same mistake. 47:22 But we also have another one, a security agent. 47:26 They'll help you build more apps, be more secure. 47:30 They'll check for permissions in Forks of Standards 47:32 and look for any loopholes. 47:35 In this example, you can see it found 47:37 a pretty significant vulnerability. 47:40 After it reviewed the app, it found 47:42 there was a hard-coded API key and then stored it securely. 47:47 We believe this is so amazing for your builders, 47:49 because this is how you can build fast and securely. 47:54 But that's not all. 47:55 We also have a debugging agent. 47:58 You can see here, it found some pretty significant errors 48:01 in a couple of queries. 48:03 And it changed how you view the data in this app. 48:06 These are the kind of things that we 48:08 think you can use to really build great applications 48:10 on Retool. 48:12 And so expert agents means fewer outages, stronger 48:16 compliance, and just more peace of mind, 48:19 because everything you build is going to be enterprise grade. 48:25 But even when you have expert agents in place, 48:28 you still need to know that your apps work end to end, 48:30 not just piece by piece. 48:33 Today, we hear from you that testing is painful 48:36 because it's done manually, and it takes forever. 48:41 But like I said before, we believe the platform 48:44 should do it for you. 48:45 Builders need automated capabilities 48:48 to remove this burden of manual testing. 48:51 So that's exactly what our new built-in native testing will do. 48:55 It's going to live right here in the app builder experience. 48:58 And creating a test is as simple as just using an app 49:02 as a user would. 49:03 And then as you use it, tests are automatically created. 49:06 And you can see when things will pass or fail. 49:10 We want to take the pain out of testing 49:12 and give you the confidence that everything you're building 49:14 is enterprise ready. 49:18 But finally, even if you built the perfect app, 49:21 tested it the early, it starts to lift somewhere 49:24 in production. 49:25 And all too often, we hear that, well, hosting is where 49:28 sometimes you have the biggest challenges. 49:32 Now, today, many of you self-host on retool. 49:36 Why is that? 49:38 Because it gives you the flexibility 49:39 to connect to your data. 49:41 It gives you power. 49:42 It gives you sovereignty and control. 49:45 And it's just plain convenient. 49:48 But you also miss out on the benefits of the cloud. 49:52 You have to manually upgrade. 49:55 You have to manually scale. 49:56 And it just takes time and resources. 50:00 That's why today, I'm so thrilled to announce 50:05 general availability of our brand new self-hosted, 50:09 but retool-managed deployment option. 50:13 Many of you require self-hosting, 50:15 but you want the convenience of the cloud. 50:16 So that's exactly what this does. 50:18 All you have to do is simply deploy a retool 50:23 in your private cloud. 50:25 And we'll manage everything. 50:27 We'll manage the infra. 50:28 We'll manage the scaling, the upgrades. 50:31 Your data never leaves your environment. 50:34 your data is secure and compliant so that you 50:37 have the ultimate flexibility and ease 50:39 to run Retool your way. 50:41 The team has been working on this for over a year, 50:43 and so we're so excited that you can finally 50:44 start using it. 50:46 We're starting with AWS starting today. 50:48 We'll be expanding to more cloud providers very soon. 50:55 So that's how we think we'll help you manage confidently. 50:58 Expert agents, native testing, and a brand new hosting option. 51:09 So we talked about building, scaling, managing. 51:14 But we all know that individual apps are just 51:17 one piece of a larger business process. 51:21 And the real transformation happens 51:23 when we need to start automating problems in a business. 51:30 And most business software isn't just about apps. 51:34 W. Edward Deming, the father of quality management, 51:37 said, if you can't describe what you're doing as a process, 51:41 well, you don't know what you're doing. 51:44 We believe a business is a collection 51:46 of 1,000 untold processes. 51:50 And this is where we believe AI will have its greatest impact. 51:54 one is deeply embedded in your business processes. 52:00 In fact, this is why we originally built Retool Workflows. 52:04 It's a visual automation engine that 52:05 helps you orchestrate the processes that 52:07 run behind the scenes. 52:10 And you can think of workflows as the nervous system 52:13 for internal software. 52:15 You can move data-booming systems, 52:17 stitch logic across your systems, trigger actions. 52:23 And because it shares the same resources as apps 52:29 across-- 52:30 for your enterprise and for security, 52:33 every workflow is ready to go, secured by default. 52:37 Today, over 10 and 1/2 million workflows 52:40 run on Retool every single day. 52:43 This is where we see apps and automation 52:46 starting to converge on a single platform 52:48 for your business processes. 52:53 Now, to show you what this looks like at an incredible scale, 52:58 I'm going to invite a very special guest to the audience, 53:00 or to the stage. 53:02 If you've ever had anything delivered from Apple, 53:06 Domino's, Tesco, and you've had delivered on the same day, 53:11 this is a guy whose platform makes it happen. 53:15 And this platform handles millions of transactions. 53:18 There's absolute zero margin for error. 53:22 And it shows the gigantic scale that Retool can support. 53:27 So please join me in welcoming Berger Ebersen from Uber. 53:31 Berger, take it away. 53:32 Please welcome to the stage Berger Ebersen, Uber, 53:37 Global Head of Partner Engineering. 53:38 [MUSIC PLAYING] 53:46 Hey, everyone. 53:47 Thank you, Berger from Uber. 53:52 And I'm not going to talk too much about AI today, 53:56 but that is an AI generated image, just to start off with. 53:59 So just saying. 54:02 Yeah, we've been using Retool for basically enabling 54:06 our partner engineers to do more integration work 54:09 and connect with partner systems all over the world. 54:13 You mentioned a couple of those. 54:14 And I'm excited to say we've managed over 1,500 applications 54:18 in our team. 54:19 Not all of them on Retool though, but getting there. 54:22 My boss actually said it based on he said 54:24 that we are one of the few companies that actually 54:26 have APIs that move people and things around. 54:30 So it's kind of cool to think about how 54:31 we can control the movement of actual physical beings 54:35 through our APIs. 54:36 And Retool has really made it easy for us to do that at scale. 54:40 Yeah, we also created this whole team and organization 54:45 around it and a program called BBU, built by Uber, 54:48 because now we've empowered our own team 54:50 to build this ourselves. 54:51 Previously, we were dependent on third-body contractors, 54:54 technical teams from partners, or just 54:57 kind of like core engineering teams 54:59 who didn't have the same priorities as maybe we had 55:02 in our partners' heads. 55:03 So Retutl has really allowed us to kind of move into that space 55:06 and create this program of BBU. 55:09 We have also a cool slogan for it, which is called JFBI. 55:12 And I'll let you figure out what that means as I go along. 55:17 Yeah, so putting the engineering back in partner engineering, 55:20 that is literally important to me because we face partners, 55:26 and that's an important part. 55:28 We need to make sure that all of Uber's partners are happy, 55:31 the technology works for every delivery, 55:33 for every transaction. 55:36 And we are partner-facing, so we get the heat when things go 55:39 wrong, and we get the heat when things move slowly. 55:42 So what the RETULE allows us to do is really just 55:46 do engineering again. 55:47 So you've heard a lot about this new breed of builder, 55:50 this new type of person who can actually kind of move things 55:53 and do it by themselves. 55:55 And Retool has really helped us to unlock that. 55:58 So we obviously have taken a very serious approach 56:01 by standardizing a lot of it, creating proper change control 56:04 processes around it, ensuring that we do things 56:06 in compliant ways. 56:07 And that is part of the program. 56:10 But Retool really makes that a lot easier. 56:12 And I'm super excited to hear about some of the stuff 56:14 that they just announced, because that'll 56:15 make it even more easy for us to kind of manage this at scale. 56:19 Yeah, I think the big point here is that Retool really 56:24 empowers us. 56:25 And the point is that technology or integration 56:29 can no longer be the blocker to a business deal. 56:32 There's no opportunity where we can say, 56:33 we cannot have this partnership or the sync on go live, 56:36 because we have some sort of tech blocker or delay 56:38 and integration. 56:40 So that's really one of the key points 56:41 that Retool allows us to celebrate every day. 56:45 We also, obviously, at Uber scale, reliability 56:50 and observability and those things are really important to us. 56:52 So we make sure that we maintain four nines, at least. 56:57 In some cases, five nines. 56:59 And Retool has given us commitments 57:01 to be able to advertise that. 57:05 So when we say our integration layer is available, 57:08 it's available all the time, not just sometimes. 57:11 And the volumes that we are starting to put through there 57:13 are quite significant. 57:15 And I'm really excited to see just how the system is holding up. 57:18 Obviously, I don't want to say anything to jinx us, 57:20 but we haven't had any major issues 57:22 since we started using Retool. 57:24 And hopefully, that continues to be the case. 57:27 So yeah, I think the other thing that Retool allowed us to do 57:31 is really just create quality programs around our integration. 57:35 So now that we control the integrations, 57:37 we can really see what the features are doing, 57:40 how it's performing. 57:41 So when we see small things that aren't working even 57:44 operationally in the delivery world, 57:46 it's really just logistics, right? 57:47 Getting people to do things at the right times. 57:51 And sometimes you can tweak small things in API behavior 57:54 and make big differences in the logistics flow. 57:57 So with Retool, it really allows us 57:59 to kind of go and tweak minuscule things that 58:02 have big impact in the larger program. 58:05 And then the last thing I'll mention 58:07 before I get into some of the logos 58:10 is that we created a tool that actually allows us 58:14 to certify our integrations automatically using AI. 58:17 So our partners go through a process 58:20 where they integrate sometimes not Uber built, 58:22 but partner built. 58:24 But we've created a program on retool 58:25 that allows us to kind of validate 58:27 what that application is doing, look at our API behavior, 58:30 and actually tell them whether or not 58:32 they've implemented the APIs correctly. 58:34 And if they didn't, we even give them feedback 58:37 on what they should change. 58:38 We can even generate code samples for them 58:40 and say, this is like how you should be adopting that endpoint 58:43 or how you should be using that object. 58:45 So it's really helped us automate our certification. 58:47 So when an integration goes live, 58:49 we basically know it's at 100% quality 58:51 and it does everything we expect it to do. 58:54 And that's all powered by Retool. 58:57 All right. 58:58 So a bunch of cool logos on here. 59:02 Obviously, very exciting team that we're part of 59:08 because we work with these partner teams every day. 59:10 These are Fortune 50 companies. 59:13 They are amazing technology companies, 59:15 and they are really demanding. 59:18 Not in a bad way, but they expect a lot from us. 59:21 And so Uber expects the same from our partners. 59:25 So some of the cool examples here on the integration side, 59:28 on the left-hand side of the screen, 59:29 is what we wanted RETOO4. 59:31 We wanted Powerback. 59:33 We wanted to build these integrations ourselves. 59:35 And we've done a bunch of that already. 59:37 And that kind of works exactly the way we wanted to work. 59:40 Very excited about the McDonald's launch, which is imminent. 59:43 And we're going to put significant volume 59:46 through that, hopefully very soon. 59:48 But the right-hand side of the screen 59:50 was kind of like a surprise. 59:51 It's something we didn't really anticipate. 59:53 But when we started using Retool, we 59:54 realized how we can create little plugins. 59:58 And they called it Hacks before, but I told them to rebrand 1:00:00 to plugins because, generally speaking, 1:00:04 people don't like to hear Hacks. 1:00:06 So we're in the plug-ins. 1:00:08 But really what these things are is-- 1:00:10 I mentioned before, those specific operational matrix, 1:00:13 you see a courier arriving too early, too frequently. 1:00:16 And they stand at the restaurant, 1:00:18 and they wait for your food to be ready. 1:00:20 And we basically realize that the stores don't 1:00:22 have good quality data of how long 1:00:25 do we need to give the courier before they get there. 1:00:27 And we have better metrics on Uberside. 1:00:29 So we can actually change and modify those transactions 1:00:32 by doing API calls and updating timestamps 1:00:35 and doing small tweaks on the transaction level, 1:00:39 which then just makes the entire experience a lot better 1:00:43 overall. 1:00:44 So the whole right inside of the screen 1:00:46 is like things that we built with Retool 1:00:47 that we didn't even expect that we were going to build. 1:00:50 And to be honest, we've built, I think, 35 of those 1:00:52 in the last year. 1:00:54 And a lot of them have very, very significant dollar 1:00:56 value and impact across the world. 1:01:00 Yeah, and even-- I mean, this Olive Garden one 1:01:02 an interesting example just because it was a big launch. Uber Direct was a big exclusive 1:01:09 partnership they announced a couple of months ago and we've been doing a lot of Olive Garden 1:01:14 deliveries. But it was a very complex project and a very hands-on project and we realized 1:01:20 we don't really have a good way to communicate with them because they use SharePoint and 1:01:24 they're Microsoft shop and we don't use SharePoint and so we just built a retool dash to manage 1:01:28 the entire project and do the validation of the API integrations, do the quality, and show 1:01:33 them the metrics of how their API is behaving or their integrations behaving, and we did 1:01:37 it all in Retool as well. 1:01:39 So it really just helped us gain that extra bit of trust and efficiency within the project. 1:01:49 So a couple more examples, and I just wanted to call these out. 1:01:52 A name dropped a few of my team members. 1:01:54 I hope they see this if they're out there. 1:01:56 Again, these aren't back-end engineers. 1:01:58 These are partner engineers, some of them 1:02:00 with varying skill sets. 1:02:03 Some of them are good engineers, so I just 1:02:05 want to give them credit. 1:02:06 Sorry, guys. 1:02:07 Some of you are really good engineers. 1:02:09 But we were able to do a lot of cool things. 1:02:11 And I love the quote from one of our operations leads there 1:02:14 in the UK. 1:02:16 We were launching with Tesco, and we realized really 1:02:19 late in the last minute that some of the things weren't working 1:02:22 the way we wanted to. 1:02:23 And the team were able to build a plugin that 1:02:27 allowed us to modify the transactions 1:02:29 and get the results we wanted. 1:02:30 And we could then scale and launch Tesco as intended. 1:02:33 And it's a very big business opportunity for us. 1:02:36 There we were able to unlock with Retool very, very last minute. 1:02:40 And yeah, I think some of the obvious things that people 1:02:43 probably do, but for us it wasn't an obvious thing either, 1:02:46 was this like a technical support dashboard. 1:02:48 So we have these large partners, point of sale companies 1:02:51 like Toast. 1:02:52 And basically, they have a large number of support 1:02:55 tickets that they get from merchants and say, 1:02:57 I worked all night with my mom to update our pizza menu. 1:03:01 And I put it on the POAs, but it's not on UberEats. 1:03:03 What's going on? 1:03:04 And we need that. 1:03:05 And so they used to then create the ticket with Uber. 1:03:08 I asked them to force a refresh of the menu, 1:03:09 get the menu menu for the POAs. 1:03:11 So we just kind of gave them a tool 1:03:13 and said, next time you get a request like that, 1:03:15 because it's the majority of your requests, 1:03:16 just hit this button on this dashboard. 1:03:18 And we've reduced our subcourt tickets 1:03:19 from, for this type of case, by like 90%. 1:03:22 So really cool things that you can just do. 1:03:26 Again, not something we anticipated. 1:03:28 Coming out of Retool as a value add, 1:03:30 we were just going to build middleware integrations 1:03:32 and an ETL layer using workflows. 1:03:34 And I'm excited about this Apple one. 1:03:37 So they came to us and said, they 1:03:39 don't like the way our standard Uber bar code works. 1:03:41 So they can't scan it in store. 1:03:42 They wanted a custom type of QR code. 1:03:46 And we use Retool to actually generate 1:03:48 like this custom label for Apple that they can then stick on their boxes and when careers 1:03:53 pick it up and they can scan it and their systems recognize it. 1:03:55 So this is something that would, on a logistics, if you ever doubt with these things, they 1:04:00 take months or years to develop different types of label scans. 1:04:03 So we were able to do this in less than a week. 1:04:05 So pretty exciting stuff for us. 1:04:09 To be honest, Shopify, everybody, there's probably a lot of people here that know and 1:04:12 and use Shopify. 1:04:13 And that's something that we built the back end on RETU. 1:04:18 Again, we didn't really anticipate it. 1:04:19 We tried to build it in a different way before. 1:04:22 It was kind of in the early days of our RETU. 1:04:23 And we just literally lost a minute, decided 1:04:26 RETU can do this better. 1:04:27 And we rebuilt the whole back end in two weeks 1:04:29 for the entire Shopify integration. 1:04:31 So it allows us to kind of pivot and realize 1:04:34 we had issues with the previous choice of architecture. 1:04:37 And RETU kind of saved the day. 1:04:41 Yeah, I think that's good enough. 1:04:45 I think the next slide, I want to just kind of go back 1:04:50 to a couple of the core principles, like really what 1:04:54 Retool allows us to do. 1:04:55 So we sit in a unique position in front of these partners. 1:04:59 We get first-hand feedback from them. 1:05:01 Our core product is amazing, and Uber 1:05:03 does volumes that are staggering. 1:05:06 But we can iterate and experiment real quickly in Retool, 1:05:10 And then we can provide that like the results back 1:05:13 to our product teams as like a formal, 1:05:15 like every half or a quarter or whatever they do they planning. 1:05:18 And then we can tell them like, look, this worked really well. 1:05:20 This drove specific metrics. 1:05:23 And we were able to kind of like succeed in these areas. 1:05:25 So we recommend that you invest in the core product 1:05:27 in a similar thing. 1:05:29 So it really helped us to kind of like not just 1:05:31 say the partner says they want something, 1:05:33 but we can say the partner said they wanted something. 1:05:35 We tried it out. 1:05:36 It worked. 1:05:37 And it should be part of the core product. 1:05:39 And that's really been a core part for us, 1:05:41 like building trust with our product teams 1:05:43 and our engineering teams. 1:05:44 And sometimes they come to us for suggestions, right? 1:05:47 Like, can you guys try something out? 1:05:49 Can we work together? 1:05:50 And you can do it on Retool first, 1:05:52 and then we see what the results are. 1:05:54 So it's really allowed us to kind of gain 1:05:57 next level of trust within the organization. 1:06:01 Yeah, I think that's about it. 1:06:02 Thanks, everyone. 1:06:03 I appreciate it. 1:06:05 Welcome to this game, Eric Chang, for Matsu Australia. 1:06:23 - It's such a fast time period with the partner 1:06:25 in cheering team and to support some of the largest brands. 1:06:28 So really awesome, thank you again for joining. 1:06:31 Now, one of the things we've talked about is this automation software. 1:06:40 Business process automation software is really great for solving valuable use cases. 1:06:47 But it's deterministic. 1:06:50 It's process-oriented. 1:06:52 It's step-by-step. 1:06:53 And there's this inspredictable. 1:06:55 There's a limit to what it can accomplish. 1:07:01 You've seen how AI has changed everything, especially when 1:07:04 applied to automation software. 1:07:07 Software doesn't look like software anymore. 1:07:09 LLMs are doing these almost human-like creative things 1:07:14 that just weren't conceivable of. 1:07:16 But due to the nature of LLMs, they're non-deterministic. 1:07:21 They can be unpredictable sometimes. 1:07:23 It can be a black box. 1:07:26 And sometimes it can be a little risky 1:07:28 to use non-deterministic solutions 1:07:31 to solve a business process. 1:07:33 For example, I'd be quite hesitant to let 1:07:37 an LLM manage my annual employee performance review cycle. 1:07:43 But what if you combined them? 1:07:46 What if you took the creative power of AI 1:07:48 with the reliability of rule-based software 1:07:51 with just a touch of human judgment? 1:07:54 We think that would change the game for enterprise software. 1:07:57 Because now, you can start performing human-like work 1:08:01 at inhuman scales. 1:08:03 That's what we believe the future of automation software 1:08:05 is going to look like, and what we believe the future of how 1:08:08 businesses will operate. 1:08:11 So this is Retool Agents. 1:08:16 Earlier this year, we launched Retool Agents, 1:08:20 a full-stack toolkit for building, deploying, 1:08:22 and managing AI agents across your enterprise systems. 1:08:27 But what's made them uniquely powerful in Retool 1:08:30 is that you can actually build agents with the actual tools 1:08:34 your business needs to get the work done. 1:08:37 And then our agents will decide for themselves 1:08:40 what tools they need. 1:08:43 So let me show you one in action. 1:08:47 Here, you can see a dispute resolution agent in action. 1:08:51 It's using a variety of tool calls. 1:08:53 What it's doing is it's accessing open dispute data 1:08:56 from Stripe. 1:08:58 It's then going to start getting into CRM data 1:09:00 from Salesforce. 1:09:02 It's then going to query data from our own database. 1:09:06 It's getting all open tickets. 1:09:08 And eventually, you'll see all of it 1:09:10 gets packaged into a submission back to Stripe. 1:09:14 And then I also get an email. 1:09:16 It's awesome. 1:09:18 But what's really cool about this 1:09:19 is we didn't tell it to solve the problem. 1:09:22 We told it how to handle the charge back. 1:09:26 And it decided what tools it needed for the job. 1:09:30 And this is just one agent, combining Stripe, Salesforce, 1:09:34 your database, and email. 1:09:37 But the bigger point here is that any business context 1:09:40 that you've exposed in Retool will actually 1:09:42 be able to be used by your agents. 1:09:44 And so the possibilities for composition are endless. 1:09:50 And we're so excited to do the attraction 1:09:52 that we have seen with Retool Agents since launch. 1:09:56 It's been less than four months, but we have over 10,000 1:09:59 agents built and executed, powering 1:10:02 so many mission-critical processes. 1:10:05 And the breadth of use cases that is supporting is incredible. 1:10:09 For example, we're seeing customer support, claims 1:10:14 processing, data analysis, data enrichment, research, sales 1:10:19 outreach, even chat box, taking action on real business data. 1:10:25 It's inspiring to see what agents have been 1:10:27 able to accomplish for our customers. 1:10:30 And the impact has also been tangible. 1:10:33 We see teams now are saving time and costs. 1:10:38 They're getting more productivity, more throughput. 1:10:42 And they're uncovering lots of revenue opportunities. 1:10:46 That's the growth we see in businesses. 1:10:50 Now, I want to invite someone to show you a real-world example of agents in action. 1:10:57 Komatsu is one of the world's oldest and largest manufacturers of construction equipment. 1:11:05 They have over 60,000 employees globally. 1:11:08 And our speaker just dropped us on us yesterday. 1:11:13 He just won the global hackathon using Retool at Komatsu. 1:11:18 So please join me in inviting Eric Cheng to the stage from Komatsu. 1:11:23 Please welcome to the stage Eric Cheng, Komatsu Australia, Enterprise Architect. 1:11:36 Thank you Abashak. 1:11:40 Good morning everyone. 1:11:41 It's a pleasure to be in San Francisco 1:11:44 and to be able to be at this summit today 1:11:46 and to share with you some of the work 1:11:48 that the team back home at Australia 1:11:50 has been working on over the past few months. 1:11:54 So we're very fortunate to have Retool 1:11:57 as a strategic technology partner to come on this journey 1:12:00 with us. 1:12:01 And the journey that I want to share with everyone today 1:12:04 is how we at Kamatsu Australia has been leveraging 1:12:07 this whole Retool platform, the application, the tooling, 1:12:10 the AI agentic layer. 1:12:12 And not just to build apps and workflows, 1:12:14 but really redefine how our customer support center operates 1:12:18 and deliver outcomes for our customers, 1:12:20 as well as how our employees within the call center do work. 1:12:26 So our Kamatsu support center, or the KCSE for short, 1:12:30 is really the face of our organization. 1:12:32 They operate 24 hours a day, seven days a week. 1:12:36 And we have about 40 employees in that support center 1:12:39 that serve our customers across Australia, New Zealand, New 1:12:43 Caledonia, and also Papua New Guinea. 1:12:46 So they operate in a traditional core center 1:12:49 sense. 1:12:49 So we have support tickets coming in 1:12:52 across emails and phone calls. 1:12:54 So about 9,000 phone calls, 8,000 emails per month. 1:12:58 And these requests can pretty much 1:12:59 range from everything, right? 1:13:00 So letting our customers know when 1:13:03 they're going to receive their order, 1:13:05 a customer asking about the service history 1:13:07 or previous service report, or even maybe a parts diagram 1:13:11 so they can actually install their part. 1:13:13 We also perform a lot of administrative functions 1:13:15 for our customers and internal branch network as well. 1:13:18 So raising quotes, sales orders, processing purchase orders 1:13:21 in our ERP so we can get paid, as well as helping our customers 1:13:25 perform locks on or geo-fence on their machines 1:13:28 in our contracts telematic system. 1:13:32 Recently, we also added to the remit 1:13:34 of our core center outbound sales. 1:13:36 So they also now have a sales target. 1:13:39 So as you can see, there's a lot of activity, 1:13:41 a lot of customer interaction that 1:13:43 runs through our support center. 1:13:45 And as we all know, there's a lot of complexity. 1:13:48 And as we all know, with complexity also comes challenges. 1:13:52 Because our core center has operated as it has now 1:13:55 for many, many years now. 1:13:57 And it's worked well in the past. 1:13:59 But like many organizations, and with all these disruptive 1:14:03 technologies, and as our organization starts 1:14:06 continues to grow, we're starting to encounter challenges and points of friction which we 1:14:11 want to overcome. 1:14:13 From an operational perspective, Kamatsu Australia continues to grow year on year, we sell more 1:14:18 machines and as a direct result, our aftermarket business continues to grow for parts and services, 1:14:23 so it's a great problem to have. 1:14:26 But as a Japanese organisation, we also ask ourselves, right, how do we balance this growth 1:14:31 sustainably, how do we manage capacity and also headcount? And if we do hire, how do 1:14:38 we actually get someone to be trained up and productive? Because currently it takes about 1:14:43 six months for that person to be productive within that sensor. And there's a number of 1:14:48 reasons for this. There's a lot of complex processes, we have regional, we have global 1:14:53 systems that they have to navigate. So it's quite convoluted. 1:14:58 as quality challenges, also goes hand in glove 1:15:00 with operational challenges. 1:15:03 So currently, we have an average handling time 1:15:06 of five minutes per core. 1:15:08 And this is also something that we track in our customer NPS 1:15:11 survey. 1:15:12 So for us, it's not really just about maintaining 1:15:15 the customer level or level of customer support 1:15:18 in the face of increasing demand. 1:15:20 It's actually how do we continue to improve it. 1:15:24 Sales performance. 1:15:26 In the midst of everything I talked about, 1:15:27 How do we find a capacity for our employees 1:15:30 to actually, due to research on our customers, 1:15:34 identify sales opportunity, make those calls, 1:15:36 and reduce customer churn and generate revenue? 1:15:39 And then lastly, as a business unit, as an organization, 1:15:42 how do we take advantage of these technological shifts? 1:15:46 How do we leverage AI, but in a secure and also 1:15:49 sustainable manner, and also in a way that also 1:15:51 aligns with our people and processes? 1:15:54 So as you can see, there's a lot of challenges 1:15:57 that we face and we need to overcome. 1:16:00 But with challenges also comes opportunity. 1:16:03 And this is where Retool really comes into the picture for us, 1:16:06 because it's provided a holistic platform that 1:16:09 allowed our organization or an enterprise to really 1:16:12 operate at a startup level. 1:16:14 We were really surprised at how quickly we 1:16:16 were able to get Retool up and running, 1:16:19 and then also to start developing our autonomous customer 1:16:23 support agent POC back in April. 1:16:26 the team at that time was completely new to Retool. 1:16:30 And in a couple of weeks, and with the support of Retool, 1:16:33 we were able to develop and complete 1:16:35 our POC for autonomous agents. 1:16:38 And as mentioned before by Abishak, in July, 1:16:42 we were fortunate enough to be participating 1:16:44 in our Komatsu Global Hackathon, fortunate enough to win it. 1:16:48 But I think what it provided us was validation. 1:16:51 Validation in our use case, but also validation in that Retool 1:16:54 was the right platform for us. 1:16:57 So I'm glad to say that over the past few months, 1:17:00 since July, since the hackathon, our team 1:17:03 has been busy working behind the scenes, 1:17:05 productionizing this whole scope of work. 1:17:07 In the next couple of weeks, we'll 1:17:09 go live with phase one of our agented customer care project. 1:17:13 And this will deliver significant gains back 1:17:15 to our business by providing customer and machine 360 1:17:20 to all our frontline staff, reducing the time 1:17:24 to deliver insights to those agents, human agents, 1:17:27 from minutes to seconds. 1:17:30 By the end of the year, we'll also 1:17:32 deliver and productionize our autonomous agents. 1:17:35 And we'll be able to close the loop 1:17:37 on improving our customer experience 1:17:40 for those email support queries. 1:17:43 And then lastly, in the next year, 1:17:45 in a medium term, with these applications, 1:17:48 with the right tools, help, and support, 1:17:51 We hope to provide a new operating model for our support sensor and to be able to make 1:17:55 our people more effective sales agents. 1:17:58 Refocusing them from doing that reactive mundane support into revenue-generating proactive activity. 1:18:09 So from a solution overview, we have the intake funnel on the left. 1:18:13 So again, as I mentioned, support tickets come in from emails, from phone calls. 1:18:18 What I didn't mention was our support center relied on a lot of Excel spreadsheets and legacy 1:18:25 systems. 1:18:26 Retool provided an opportunity for us to move away from Excel, modernise those applications 1:18:33 as well as those processes as part of process re-engineering. 1:18:37 The agentic layer is based on or built on retail agents. 1:18:41 And a significant of that is that we don't need to actually worry about orchestration, 1:18:46 about the monitoring, about the memory, about evals. 1:18:49 Retool provides everything out of the box for us 1:18:52 so that our team could really just focus on delivering 1:18:55 the outcome for our business. 1:18:58 We also delivered an AI gateway in the tooling. 1:19:02 So the tooling are basically the arms and legs of our agents. 1:19:05 It's all built on Retool workflows. 1:19:07 And these Retool workflows connect 1:19:10 to our various back-end systems, applications, databases. 1:19:15 And they basically surface these information back to an LOM, 1:19:19 back to our human agents. 1:19:21 On the topic of human agents, there's always a human 1:19:25 in the loop. 1:19:26 So they're going to be at the end of every customer interaction, 1:19:30 making sure that they have oversight of all our customer 1:19:36 communication. 1:19:40 So once we deliver this whole program of work in full, 1:19:43 The value proposition for us is quite significant. 1:19:46 Over 22,000 hours of efficiency gain 1:19:48 in terms of productivity, in terms of modernization 1:19:54 and process engineering. 1:19:56 We'll be able to refocus our customer support agents 1:20:00 to revenue-generating work and be 1:20:02 able to contribute to that 8 million target I 1:20:05 talked about earlier. 1:20:06 We'll be able to reduce by 30% to 40% 1:20:10 of our average handling time for phone calls, 1:20:12 as well as our meantime to resolution for emails. 1:20:15 And that will improve our customer experience. 1:20:17 But I think the most significant for us 1:20:19 is the transformation aspect, not just for us 1:20:22 as a business unit or Commatio Australia, but globally. 1:20:25 So I'm glad to say that as part of that hackathon, 1:20:28 it generated a lot of interest for retail. 1:20:31 And as part of this AI working group that we have globally, 1:20:35 we're now-- 1:20:35 Commatio Australia is now leading the charge 1:20:38 in terms of developing a proposal for a low-code application 1:20:42 development platform. 1:20:43 So it's quite exciting there. 1:20:45 So in summary, how has Retool enabled Kamatsu? 1:20:49 Retool's provided a holistic platform 1:20:52 that provides the application, the workflow, the AI, 1:20:55 Gentik layer. 1:20:56 It's provided the guardrails that 1:20:59 allowed us as an organization to really 1:21:02 operate at a startup level, accelerate the development 1:21:05 and time to value for our business, 1:21:08 reduce the cost and risks for working with AI. 1:21:12 And that's quite significant for us, 1:21:14 because we have a relatively small team. 1:21:18 And I think also, more significantly, 1:21:21 we feel that Retool will set us up in the future 1:21:25 to meet the demands of the organization, 1:21:26 regardless of any shifts in technology. 1:21:30 Thank you. 1:21:31 So hand us back to Abhishek. 1:21:33 [APPLAUSE] 1:21:37 [APPLAUSE] 1:21:40 Thank you, Eric. 1:21:43 It's amazing to hear what they've accomplished on Agents. 1:21:46 It's three weeks to create a POC. 1:21:49 And Komatsu has shown the power of Agents 1:21:51 making an impact today. 1:21:53 It's inspiring to see all that they have accomplished 1:21:56 in such a short period. 1:21:58 And there's so much we can all learn from them 1:22:00 in how to truly use AI to transform your businesses. 1:22:06 And when we talk about using the added transform businesses, we talk about agents and the impact 1:22:10 that we see with agents, we believe this is what the future of internal software is going 1:22:14 to look like. 1:22:17 Agents and enterprise software is the majority of the type of software we'll see is going 1:22:22 to get created over the next few years. 1:22:27 So there you have it. 1:22:29 We love internal software. 1:22:31 We're excited to build with you. 1:22:33 We're excited because this is a software that 1:22:35 makes the world go around. 1:22:37 We want to help you build, scale, manage, and automate. 1:22:41 Our roadmap was built for you. 1:22:45 Now, before I go, I want to raise a toast to all of you, 1:22:50 the builders, the people who keep the world's 1:22:53 businesses running every single day. 1:22:56 Our mission has always been about bringing 1:22:59 good software to everyone. 1:23:01 We want to be a platform that allows 1:23:03 to move faster, build safely, and bring innovation 1:23:07 to where matters the most. 1:23:10 So thank you all for listening. 1:23:11 And I hope you enjoy the rest of the show. 1:23:13 Thank you. 1:23:15 [APPLAUSE] 1:23:16 [BLANK_AUDIO]
