0:03 Please welcome retool product marketing 0:07 Todd Powleti. 0:15 Thank you. 0:18 Hey everybody, 0:20 welcome back from the break. Uh, I was 0:23 told the energy might be drifting a 0:25 little bit, so I thought I might do some 0:26 swearing like, uh, David did this 0:28 morning, but I don't think I'll do that. 0:31 Um, as the voice said, my name is Todd. 0:33 I'm in product marketing here. I'm going 0:34 to be joined in a few minutes by 0:36 Gabriella. She's our lead product 0:38 manager over the assist functionality 0:40 that we've been talking about throughout 0:42 the day and launched this morning and a 0:43 lot of the AI work that we put into the 0:45 platform uh, over the course of the 0:47 year. You probably all know Gabriella. 0:49 She spent a lot of time talking with you 0:51 and it's nice to see some of you guys 0:53 that I uh recognize as well. We want to 0:56 talk today in this session about how to 0:59 safely democratize development. This new 1:02 class of builders and now uh throughout 1:04 the organization 1:06 um tomorrow's builders we referring to 1:08 them as how do we let them build but do 1:10 it without creating chaos 1:13 without stressing ourselves out or the 1:15 company out and create without adding 1:18 risk. AI puts us in a phenomenal 1:20 position to do this with appgen and 1:24 codegen tools. It puts us on the 1:26 precipice of what could be the most 1:30 efficient era of application development 1:32 and application generation that we've 1:34 ever lived through or it puts us on the 1:37 precipice of 1:40 potentially the biggest mountain of tech 1:42 debt that the world has ever seen. We've 1:45 been optimizing the product for the 1:47 former and we want to make sure that we 1:49 can avoid the latter. 1:53 So when we spent some time talking to 1:56 most and many of you over the last 1:58 couple of quarters to kind of get a 1:59 sense from you how you wanted us to 2:01 incorporate uh AI application generation 2:05 into the rest of retool. We did want to 2:09 spend some time kind of botting them out 2:10 botting them out some problems that we 2:12 were actually solving. And we weren't 2:14 just adding this functionality for the 2:15 sake of adding it. What are we trying to 2:17 actually accomplish? And we ran into 2:19 some pretty common themes. These might 2:22 be familiar to you. We've been working 2:24 on them uh um you know with with you 2:27 guys for the last five or six years. And 2:29 it was a bit encouraging and 2:30 discouraging. The main theme and we've 2:32 basically captured it as the enterprise 2:35 developers dilemma. And the dilemma is 2:38 about choice. How much control and 2:41 flexibility do you put into the hands of 2:43 the business teams out in the business? 2:46 Uh as opposed to maintaining centrally 2:49 and building to support them in a 2:51 controlled, secure, easier to maintain 2:56 manner, right? Do we let the business 2:58 teams buy their own SAS applications and 3:01 rent their own s SAS applications 3:02 because it solves a particular problem 3:04 or a niche that they're that they're 3:06 working on or do we build those things 3:08 for them? 3:10 They're risks and tradeoffs. Do we let 3:13 them develop things on their own? Do we 3:14 let them use spreadsheets to accomplish 3:16 certain uh tasks and meet certain goals? 3:19 Do we put engineers and developers out 3:21 into the business in the divisions and 3:23 the departments to directly support the 3:26 teams? But if we do that, there's there 3:28 are tradeoffs, right? This is this 3:29 dilemma, this back and forth. We then 3:32 have to integrate a lot of SAS systems 3:34 that are fragmented. We have to bring 3:36 all of our data together. Uh we have to 3:39 manage against new security 3:42 vulnerabilities. Um integrate with 3:44 legacy systems. So there risks and 3:46 trade-offs, but you know, the businesses 3:49 are not slowing down. 3:52 uh you know our our our counterparts 3:54 need to continue to move fast to do 3:56 their jobs and AI is putting pressure on 4:00 this system. They all begin to want to 4:02 use some of this to build things on 4:05 their own. Um but if we hold back and 4:07 we're hand building things for them or 4:09 using uh tools as they exist today for 4:12 them, the backlog gets longer, tension 4:16 gets tighter, friction increases, and we 4:18 kind of have this this this this 4:20 standoff. 4:23 So we want to solve for this and AI puts 4:26 us in a phenomenal position to do it, 4:30 but it hasn't happened yet, right? We're 4:33 searching for this ROI. We're pretty 4:36 sure it's out there. We've seen glimpses 4:39 of it, but we haven't been able to 4:41 confirm it. There's a lot of money going 4:43 in. Billions of dollars going in. Every 4:46 day you wake up and you scroll through 4:47 the headlines and you see billions going 4:50 into data centers, into chips, into 4:53 deals, to circular deals, venture 4:55 capital being deployed, software startup 4:58 after startup after startup, billboard 5:00 after billboard after billboard that we 5:02 drove through and saw on the way here. A 5:05 lot of money going in, but we haven't 5:07 seen repetitive yield coming out, but we 5:11 know it exists. And it's models not the 5:15 problem. Some of the core tools that 5:17 we've seen so far aren't the problem. 5:18 The models are phenomenal. They're 5:19 life-changing. They will continue to be 5:21 life-changing. 5:23 Some of the code generation tools are 5:25 good. We've used them. I'm sure many of 5:28 you have used them. The appg platforms 5:31 that David mentioned earlier today that 5:33 were super popular in the beginning of 5:34 the year have trailed off in their 5:36 popularity because they're stuck in 5:39 prototype land. Right? In this phase 5:41 that we're in, the demos look amazing. 5:43 The prototypes look great. There's some 5:46 PC's that are working, but we haven't 5:48 turned that corner to have repetitive 5:52 production applications using AI to 5:55 prompt to build to build apps. And it's 5:59 because the governance systems, the 6:01 integration, the security, the guard 6:03 rails, that's the hard part, right? 6:06 That's what's great about the way that 6:08 we've been able to approach it with you 6:10 guys here at retool and taking your 6:12 feedback is we've been able to 6:14 incorporate genai application uh 6:16 generation capability into the core 6:19 retool platform which is already solved 6:21 for these things right so we're set up 6:24 for success 6:25 we have this paradoxical notion here 6:27 that we talk about which is constrain to 6:29 unconstrain so if we can put in place 6:32 the right security parameters the right 6:34 authentication 6:36 the right access controls, the right 6:38 libraries of tools that are pre-approved 6:41 and pre-built centrally and we have that 6:44 constrained then we can unconstrain the 6:47 development process and this new class 6:48 of builder can begin to participate in 6:51 building and we'll start to get to that 6:53 super efficient era uh of application 6:56 delivery that that we're hoping for. 6:59 So who are tomorrow's builders? I'm sure 7:02 you've heard us talk about them in many 7:04 stages uh uh sessions today and some of 7:06 the customers we had great sessions 7:09 covered this. 7:11 They already work at our company. That's 7:15 great. They're the operations leads, the 7:18 business teams, the data teams, the 7:20 analysts. They work in HR and finance 7:22 and revenue and the supply chain group 7:25 in sales. In large part, they're the 7:27 teams that we've been working with all 7:29 on to build internal tools for them, 7:32 right? On this 2 by two, they're high on 7:34 this side of domain knowledge, right? 7:36 They're in the business. They're close 7:39 to the problem. They know what needs to 7:41 be solved. They're the users of these 7:42 applications also, right? So, in a way, 7:46 they're primed to be efficient builders 7:50 because they know exactly what they 7:51 want. They haven't had and don't have 7:53 coding expertise. And we're not going to 7:55 try to fill that part of the box in with 7:58 natural language application 7:59 development. They don't need it as long 8:02 as we apply some of the constraint and 8:05 the guardrails so that they don't run a 8:07 muck and create this mountain of tech 8:10 debt. 8:12 They already work here. 8:16 So if these are the new builders, 8:20 what happens to the rest of us? 8:23 Tim O'Reilly had a phenomenal article 8:26 earlier this year with a provocative 8:31 headline on purpose where he's 8:33 describing the end of programming as we 8:35 know it. And it was around all this 8:37 chaos that was swirling around code 8:39 generation 8:41 tools are going to eradicate the 8:43 software development career and 8:45 practice. We don't need developers 8:48 anymore. His point wasn't that. His 8:53 point was the as we know it point that 8:56 it's the process will will evolve. We 8:59 strongly agree with him in all of our 9:02 conversations with you over the course 9:03 of the summer and building this. You 9:05 would agree with him, right? Code 9:08 generation with AI is putting pressure 9:10 on the lower end of that career ladder 9:12 for sure, but it's not going to 9:13 eradicate the practice. Right? We will 9:16 evolve. his point also in the article 9:18 which we strongly believe with. It's one 9:20 of the reasons we're in business. It's 9:21 one of the reasons we have a soldout 9:23 summit session here today and you guys 9:25 are here is that software developers 9:27 have always looked to software to help 9:29 build software. It's what retoolled is. 9:33 And so Genai is just the next phase of 9:36 that, right? It's a pretty powerful 9:38 phase for sure. 9:40 um maybe one of the most powerful that 9:42 we've seen, but it's we've seen the 9:46 movie before, right? Um so the practice 9:50 isn't going away. What'll evolve is that 9:53 we won't be at writing code by hand as 9:55 much, deploying it, managing it by hand 9:58 as much or even using some of the 10:00 classic functionality of retool. We can 10:03 rely on the AI for some of that. And so 10:06 our our role elevates a bit to not just 10:10 be handwriting code and managing 10:12 systems, managing intelligent systems, 10:16 providing access and the governance and 10:18 setting up the guard rail so that a new 10:20 class of builder the teams that are 10:23 closest to the business 10:25 can start to work on this supply chain. 10:29 And if we get those things going then we 10:32 will be in the most efficient era of 10:34 application development uh that we've 10:37 seen. 10:41 So with that Gabriella is going to come 10:44 out and she's going to get into a lot 10:46 more specificity demos and detail of 10:49 functionality 10:51 uh of the guardrail parameters that 10:53 we've built in. Um so you get to see all 10:56 of that. Gabriella, 10:58 >> please welcome product management 11:00 Gabriella Angie Olo. 11:03 [Music] 11:07 [Applause] 11:08 >> There you go. 11:10 [Music] 11:14 >> So the question becomes, how do we 11:17 empower all of these builders? 11:21 How do we operationalize it? 11:23 And so what we did is we talked to many 11:25 of you and here's what you had to say. 11:30 You came up with five things that you 11:32 would need to see 11:35 happen. The first is around data access. 11:39 You talked to us that there's two main 11:42 issues when it comes to data access. One 11:45 is that the production data for your 11:47 company could be changed without someone 11:50 knowing. 11:52 So imagine going in and seeing some of 11:54 the invoices updated. 11:57 A second is that a query 12:00 could look plausible but could actually 12:03 be wrong. And a great example came from 12:06 one of our customers 12:08 who has account data in both Salesforce 12:11 and in Snowflake. Developers know to go 12:15 into Snowflake. It's the data warehouse. 12:18 It's where the golden tables are and 12:20 they know how to look that up. But 12:22 that's a tall order for tomorrow's 12:24 developers. 12:27 And what happens is then they're using 12:29 these queries to build applications. 12:32 Applications that your business makes 12:34 decisions on. 12:40 The second thing we heard was around 12:42 reuse. 12:45 You want to make sure that reuse is part 12:48 of the system. You do not want AI or 12:51 tomorrow's builders creating the same 12:53 code over and over again. Especially 12:56 with the probabilistic nature means 12:59 there's a chance of bugs, 13:00 vulnerabilities, and again creating code 13:04 that isn't it might seem right but it's 13:07 not actually right for your company. 13:12 The third issue was around consistency, 13:15 especially with design. 13:19 We heard that there's really three main 13:22 issues when it comes to consistent 13:24 design. One is 13:27 hitting your brand standards. The second 13:30 is then you have a lot of apps and each 13:32 one of these apps have a slightly 13:34 different interaction patterns. buttons 13:36 are in different places and this means 13:38 that there's learning time for people 13:41 using these applications. There's also a 13:44 risk of them clicking in the wrong spot. 13:47 And the last is around trust. 13:50 If apps are created quickly and they 13:53 aren't looking as polished, then people 13:56 might not trust them as the official app 13:59 to do something like submit their 14:00 invoices. 14:04 The fourth concern is about governance. 14:07 A great example is the one on screen 14:09 around form validation. Engineers know 14:13 about error handling, typing, null 14:17 checks, 14:18 but often business users don't know to 14:21 do this and I doesn't reliably always do 14:25 these actions. 14:29 And the last is around safe editing. You 14:32 have apps that you've created for 14:34 months. You've really fine-tuned them 14:37 and made them work exactly as you want 14:39 for your business. And now you're going 14:41 to have a whole new population going in 14:44 and editing these apps. Typically use 14:47 get to do release management. But again, 14:50 this can get confusing really quickly 14:52 and if you don't know exactly what 14:54 you're doing, it is errorprone. 14:57 But it's truly important that there is a 15:00 way for tomorrow's developer to have an 15:03 ease of release management. 15:08 So what do you do? Do you go and say 15:13 this population can't go and use AI? 15:17 Probably good luck with that because 15:19 they're going to take out their credit 15:20 card and start using one of these other 15:22 systems. 15:24 Or do you go and lock everything down 15:26 and say anything that you create has to 15:29 go through a change that well that's not 15:32 scalable to review every change that 15:34 happens. 15:37 Instead what we've come up with you is a 15:40 governance first architecture 15:42 and the primary thinking around this 15:45 really has to come with a first 15:47 principles thinking about what AI is 15:49 good at and what we're good at. And what 15:52 we've seen is AI here is great at 15:55 generation. It can create the apps 15:56 really quickly with a lot of 15:58 functionality, but they don't know your 16:00 company. They need to partner with you 16:03 to come up with what is very specific 16:06 about your company and encode it so that 16:09 it can be reused. 16:12 So when we looked at it, we said we 16:15 don't really want someone let's say from 16:16 finance coming in and vibe coding 16:19 authentication or audit logging. I mean 16:22 I wouldn't want to vibe code 16:24 authentication from scratch without 16:26 especially not knowing if I didn't know 16:27 what I was doing. And so we think it's 16:30 better if a platform team or an 16:33 architect does this once, does it 16:35 correctly, and verifies it, and then 16:38 every app that is made by people at your 16:41 company, no matter what their skill 16:42 level is, will automatically get 16:44 authentication done correctly. And this 16:48 is what actually differentiates Assy out 16:52 there in the market against the other 16:54 products. 16:57 And tying this back to the keynote, this 16:59 is what Abashek showed you earlier 17:01 today. 17:02 And when it comes to a governance first 17:05 architecture, two parts are the key 17:09 pillars of a governance first 17:11 architecture. Manage confidently and 17:14 scale consistently. In the rest 17:17 presentation, we will go and through 17:20 each aspect of these two pillars in 17:23 detail through example. And so we'll be 17:26 going and showing you the product again 17:28 and again so you can see how this is in 17:32 action. 17:34 So let's look at scale consistently. 17:38 It is comprised of standards, reuse and 17:42 policies. 17:44 A great example of standards in retool 17:48 is theming. 17:51 Here in our theming product, 17:55 here in our theming product, you can see 17:57 all the components that come out with 17:59 retool out of the box. There are charts, 18:03 inputs, tables, and buttons. 18:08 Now, out of the box, each of these 18:10 components work fine. They're 18:13 functional, but they're kind of generic, 18:16 and that's on purpose because they 18:18 haven't had your design system applied. 18:21 Also, you haven't decided whether you 18:24 want the filter on the table to be on 18:26 the bottom right or the top left. And 18:29 this is where the theme editor comes in. 18:33 You can see on the right side 18:37 um 18:43 uh okay you can see that on the right 18:47 side as someone is using theming product 18:51 they can choose the border radius. So 18:53 the buttons on the top go from 18:55 rectangles to ovals. 18:58 Also, if someone wants to change the 19:00 primary color, they just change it here 19:02 and then it cascades to every single 19:04 component even beyond the buttons. 19:08 And the key again here is that you only 19:12 have to do this once and then it is 19:15 applied throughout your organization and 19:17 every single app builder will see in 19:20 this case for example pink buttons. 19:26 And let's look at it in practice. So 19:28 here a builder has asked for a 19:31 promotional campaigns application to be 19:34 built. Assist generates the plan, 19:38 goes and builds out some charts, a table 19:41 and even some metrics 19:45 and then it automatically applies the 19:48 finance theming. And the beauty of this 19:52 is that the builder didn't have to say 19:54 make it blue. They didn't say use this 19:57 font. They just said that they wanted 20:00 the dashboard to do and what it wanted 20:03 it to look like. 20:08 And here you can see what it looked like 20:10 before and after once the theming was 20:13 applied. 20:15 And next, let's get into reuse. There 20:18 are two parts to reuse. One is the 20:21 definition and permissions of each 20:23 element and then second is maintaining 20:27 and reusing each element. And this is 20:30 particularly helped by AI. 20:35 Here you're watching someone build a 20:37 generic function. They're an expert in 20:40 this database. It's the HR database and 20:43 they're creating a query that can go and 20:45 pull out employee data. They built this 20:48 once, tested it, and made sure that it's 20:51 exactly correct. 20:55 To enable reuse, they publish the 20:57 function. They put in the definition, 21:00 and this is helpful both for a builder 21:02 because then they know what the function 21:04 does, and then it's also helpful for AI. 21:08 They also attach it and add it to a 21:11 library as part of publishing. 21:13 And retool has a very powerful 21:17 enterprisegrade 21:19 library that it's building that will 21:21 enable you to have different sub 21:23 libraries for whatever part of your 21:25 organization you desire. For example, 21:28 you could have a sub library for 21:30 everyone in AMIA or one for people in 21:33 your HR department. It's really up to 21:35 you. And within each of these sub 21:37 libraries, you can define who can add to 21:40 the library, who can review the changes 21:43 before they're added, who can maintain 21:45 it, and then who can actually use from 21:47 each of the library. So someone that's 21:49 in the HR library and doesn't have 21:53 access to the finance one would not see 21:55 anything in finance. 21:59 So let's look at this in action. 22:01 A builder tells cyst, I want to build an 22:05 employee lookup table. Assist does not 22:08 generate this from scratch like other 22:10 solutions out there. Instead, it finds 22:14 the function, 22:16 uses it, and applies it automatically. 22:21 The builder didn't even need to know 22:23 that the function existed, which is 22:25 great. You know, you don't have to look 22:26 through the data catalog. They have to 22:28 figure out the right way to query and 22:31 they don't even need to verify it's 22:32 correct because we already know that 22:34 it's correct. They just described what 22:37 they generally wanted built and assist 22:40 found the right building block. 22:46 And as your library grows, not only is 22:49 everyone able to build faster because 22:52 now they have a bunch more Lego blocks 22:53 to pull from, but they're also going to 22:56 build correctly. 23:00 Now let's talk policies. 23:02 You can think of policies as rules. So 23:05 for example, if you have a table that 23:07 has over 100 rows have pageionation or 23:10 if you have the forums that was earlier 23:12 have validation on that. 23:16 Another example that you can see here is 23:18 a readme. The rule is that before any 23:21 published app happens, there must be an 23:23 readme generated. So as a user comes in 23:27 and says I want this app to be published 23:30 assist comes and 23:33 automatically uses AI to generate the 23:36 readme. 23:38 A huge benefit of this for tomorrow's 23:40 builder is let's say they hadn't ever 23:42 generated an app before and they didn't 23:44 know what a readme is. Now they're 23:46 learning about the readme and next time 23:48 they go into another app they're going 23:50 to know to look for this readme. 23:57 So what's happening here is that 23:59 governance is part of the building 24:01 process and it's helping guide and 24:04 educate users at the same time. 24:07 Now let's go to manage confidently the 24:10 second pillar of governance. 24:17 One of our early adopters, Samuel, he's 24:20 fantastic. He had a lot of input into 24:22 our product. He's at Felix and he talked 24:25 about how VIP coding has been great. 24:27 They've done a ton of prototypes. You've 24:29 heard about this plenty today. 24:31 But what really he was looking for for 24:34 putting applications into production was 24:38 reliability and security. 24:42 And that trust and reliability comes 24:44 from permissions that are centrally 24:47 managed. Security that's baked into your 24:50 system, a robust and easyto use release 24:53 management process, and also 24:55 observability. 24:59 So with assist, 25:02 assist doesn't have any special 25:04 permissions. 25:06 It has your permissions. So if you're 25:09 able to see a table, it's able to see a 25:11 table. And if you're able to access a 25:14 certain library, it's able to access a 25:16 certain library. 25:19 So if we look at this in practice, the 25:21 user had asked for the schema of po 25:24 Postgress 25:26 and now an admin comes in the project's 25:28 over and they remove the Postgress 25:32 visibility from the user 25:35 and just with a few clicks the user no 25:39 longer has access to the Postgress 25:42 database and that means that assist no 25:45 longer has access to to the database as 25:48 well. And so when the user comes in and 25:52 types into the system, hey, can I know 25:55 what the schema is for Postgress, you 25:58 can see that assist thinks about it a 26:00 little bit and then informs the user 26:03 that it doesn't see a Postgress 26:05 database. 26:07 That means that your permissions are at 26:10 an enterprisegrade level, no matter if 26:12 it's done by the user or it's done by AI 26:15 on behalf of the user. 26:19 Next area we're going to build upon the 26:21 last one and talk more generally about 26:23 security. And the first area is SSO 26:27 single sign on. So retool has single 26:30 sign on built in. Whenever someone needs 26:33 to log in to build, you can make sure 26:36 that they go through a single sign on 26:38 provider like octa. 26:41 What we have that's even more that's 26:44 that's differentiated is that if you go 26:47 and create applications, you can make 26:49 sure that each application also has 26:51 single sign on. So let's say that you 26:54 have a support app, anyone support 26:57 that's using that app to close out 26:58 tickets would go through whatever your 27:02 single sign on provider is first. 27:05 The second area we'll talk about is 27:07 expert agents. You saw a little bit of 27:09 that earlier today and then we will be 27:12 talking about administrative privileges 27:14 when it comes to AI. 27:18 So here is an expert agent in action. 27:21 It's our security agent. It's seen that 27:23 an API key is readily readable within 27:27 the application and that's pretty bad. 27:30 So it lets the user know and then it 27:33 collaborates with the user to remove the 27:37 API key from the app. And again, this is 27:40 another great learning moment and it 27:43 helps a user know that this is not 27:45 something you should do. And the user 27:47 could even go in and ask them why and 27:49 learn more. 27:53 We also know that at each one of your 27:55 companies, you have your own way that 27:57 you would like LLMs used. And we've 28:00 taken this in account and from the very 28:02 beginning of this product, we have 28:04 enterprisegrade controls. 28:06 You can bring your own key from any of 28:09 your providers and you can also choose 28:12 your preferred provider to work with. So 28:15 if you only want us calling anthropic or 28:17 open AI models, we will only do that. 28:22 So next comes to releases and this is 28:25 one of the most important areas that we 28:27 offer. 28:29 We know that building is mostly about 28:33 maintaining an application. In fact, our 28:35 data shows us that about 80% of all 28:39 actions done in retool are editing 28:42 existing applications. 28:44 So it's little it's like 81.5. 28:47 And with that um tra you know we talked 28:52 about before how traditional release 28:53 management gets complicated and is 28:55 errorprone and that's a lot to ask and 28:59 so we've made it really easy to have 29:02 release management in our system. 29:05 So you can see here that there's a 29:07 button called submit and there's a new 29:10 user that hasn't edited this app before. 29:12 So they want to see what previously 29:14 changes have happened. they can go into 29:16 releases and really easily revert to the 29:19 last version and see, oh, it was called 29:22 button. And while a really 29:25 self-escriptive name that makes sense in 29:27 a lot of ways, it's not a great name for 29:29 a button in an application. So, they 29:32 click within assist and can revert it 29:34 back. 29:35 You can see that assist naturally goes 29:38 in and saves the application at various 29:42 points that you can revert back easily 29:44 within your own session. 29:47 This is enabling a really easy way to 29:51 maintain the versioning of your 29:53 application. 29:55 And the next is around publishing. 29:58 We enable you to choose where you would 30:01 want your apps made in retool deployed. 30:04 You can put them on your own servers. 30:07 You can put them in your cloud or even 30:09 our cloud. It's pretty flexible. But a 30:12 builder doesn't know that. All they know 30:14 is that they want to publish their 30:16 application. And we make it extremely 30:18 easy with saying publish, which is in a 30:22 natural language that makes sense to 30:23 them. No staging and commit. 30:27 And the last is observability. 30:29 We both have the ability for you to see 30:32 a lot of analytics on the apps created. 30:35 And today we're going to dive a bit into 30:37 audit logs. 30:40 When an action is made in assist that is 30:44 part of our audit logs, you can go into 30:46 the audit logging tool and see that 30:49 assist made that change. 30:52 In addition, assist is similar um audit 30:55 logs is similar to SSO to where all apps 30:59 created can have audit logs by default. 31:05 So today you've seen many examples of a 31:09 governance first architecture where 31:11 someone has done something once at your 31:13 company and then it's automatically 31:16 applied to all applications being 31:18 created 31:20 from a builder's perspective then they 31:23 are just focused on what their 31:25 application can do not the 31:28 implementation details and so instead of 31:31 working with AI to generate code, it it 31:34 seems to them that they're more doing 31:36 logic building. 31:40 And as you invest more into the 31:42 platform, it only empowers that velocity 31:47 and doesn't constrain it. 31:51 And don't just take our word for it. 31:53 This is a really recognizable company 31:55 that has now put dozens of people onto 32:00 our AppGen product and they've been able 32:03 to build multiple applications 32:06 right into their production environment. 32:11 Those apps have been run thousands of 32:13 times a day and um they've been able to 32:17 create a bunch of value for their 32:19 company. 32:22 And this is the ultimate promise of 32:24 retool. It's the creation of what we 32:27 call the 10x organization. And this 32:30 really alludes to that kind of 10x 32:32 engineer that you've all heard about. 32:34 And imagine doing that for everyone at 32:36 your whole company. And this is where 32:38 the distance between problem 32:40 identification and production ready 32:43 solution approaches zero in the limit. 32:47 It's where you empower those closest to 32:49 the challenge to come up with the 32:52 solution. 32:54 Tools become immediately available 32:56 across your organization 32:59 and everyone can contribute to making 33:02 your company better. Thank you.
