Description Transcript
The most successful organizations won’t be those who outsource critical pieces of their business—nor those who gatekeep app development for a privileged few. It’ll be those who successfully widen the aperture of who can build software.
Product Marketing Manager Todd Paoletti and Product Manager Gabriella Angiolillo share insights into how to democratize development and enable more people to take a solution from idea to production while maintaining oversight, security, and quality. They share frameworks for who should build what, when, and how Retool makes it possible to scale impact without sacrificing control.
Recorded live at SFJAZZ Center in San Francisco on October 7, 2025.
👉 Start building in Retool for free: https://retool.com/
Read more 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.
