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]
