[RECORDING] Make Work Flow: Automation Finance presentation updates with Codex

[00:01:07,758] Kenna Valdez: Welcome to Make Work Flow.

[00:01:17,385] Kenna Valdez: This is our series focused on showcasing how real agent activators. The people who design, develop and operationalize recurring AI workflows for their teams.

[00:01:27,074] Kenna Valdez: Do just that, we'll share. How members of all kinds of different teams that OpenAI built workflows to help with real work.

[00:01:34,602] Kenna Valdez: So, whether or not you call yourself, or think of yourself as a champion of AI adoption, yet this series is really meant to give you real examples of useful workflows that you can take back and adapt to your team and tips to design.

[00:01:48,919] Kenna Valdez: Build and operationalize those workflows responsibly and effectively. Couple housekeeping items. I will say, this session is being recorded.

[00:01:57,559] Kenna Valdez: The presentation will be shared in the champion community. On OpenAI academy, where you most likely signed up to attend this session. And I'll be in chat.

[00:02:07,062] Kenna Valdez: Throughout the session today, please feel free to raise your questions as they come up. And I will try and surface as many of the top questions as we can.

[00:02:16,566] Kenna Valdez: Before the end of the session, after the presentation today, and and this particular session at Karthik, has joins us, he engineers AI success over here.

[00:02:26,007] Kenna Valdez: At OpenAI to share how Codex can help teams, automate important finance, presentation updates.

[00:02:31,808] Kenna Valdez: And we did a little bit of a bait and switch before I let Karthik introduce himself in a little bit more detail.

[00:02:38,720] Kenna Valdez: We're going to show you ChatGPT Work in Codex, which I think you're actually all really going to love. So excited to see that.

[00:02:46,495] Kenna Valdez: So, Karthik, tell the audience a little bit about you, and what problem you thought that you were. You were seeking to solve for this workflow.

[00:02:56,331] Karthik Narayanan: Hi everyone, my name is Karthik. I'm an AI success engineer here. At OpenAI. Crazy title. But what that means is, I work closely with our customers.

[00:03:04,451] Karthik Narayanan: In like regulated industries to find and deliver ways that AI can change. How their businesses work and run. So I work closely with customers and regulated spaces.

[00:03:13,394] Karthik Narayanan: So what we've been running into with my customers is they have a really large desire to use Codex. And the work product in their day-to-day workflows.

[00:03:21,624] Karthik Narayanan: But they've sometimes might run into data governance blockers that prevent them from using like connectors and other tools.

[00:03:28,373] Karthik Narayanan: So we were working primarily with finance teams, which is a near and dear issue. To my heart I started my career, an investment banking.

[00:03:35,835] Karthik Narayanan: So I've known a lot of these problems that they're facing. So I worked with a couple of customers to design a workflow that allows them to take updated financial materials.

[00:03:45,271] Karthik Narayanan: Go through and update you. Take those materials and update their models, and use those models to update their slides and presentations.

[00:03:52,678] Karthik Narayanan: That they present to leaders and share for any sort of updated communications.

[00:03:56,958] Karthik Narayanan: This is something that I wish existed when I was a young banker, but I'm excited to be able to build this with customers.

[00:04:03,597] Karthik Narayanan: And hopefully I can show you guys how we put it into production and could be a good process to share with people at your company. As well.

[00:04:11,168] Karthik Narayanan: Cool again, I'm going to go ahead and share my screen.

[00:04:16,365] Kenna Valdez: All right I will confirm for you. You are good.

[00:04:21,931] Karthik Narayanan: Amazing some small screen sharing hiccups earlier. So essentially we're going to use ChatGPT Work within the Codex app to help us update some finance presentations.

[00:04:31,099] Karthik Narayanan: We're going to have some underlying financial documents, but it's important to realize that this is a synthetic example.

[00:04:37,808] Karthik Narayanan: We're going to take a look at a prior executive readout some actuals from like financial documents that happened in may. Some supporting finance commentary.

[00:04:46,528] Karthik Narayanan: And then an updated financial model and today's goal isn't just to generate slides. Rather, it's for us to have a workflow that allows us to understand what's changed.

[00:04:55,864] Karthik Narayanan: Identify the affected slides in our presentation, preserve that source of truth, and make sure that the final presentation reconciles back to our underlying financial model.

[00:05:05,536] Karthik Narayanan: So I'll go ahead and start with the prior presentation, and I'll show you guys what we were working with. So this version was prepared before our latest financial close.

[00:05:14,983] Karthik Narayanan: It has, like some plan level performance. And the original valuation view.

[00:05:19,120] Karthik Narayanan: And instead of like manually reopening every spreadsheet that feeds into this presentation and checking every number and rewriting the commentary to make sure it aligns.

[00:05:28,568] Karthik Narayanan: We want Codex to help us step in with that part of the workflow. So we can quickly flow through and look at some of these numbers.

[00:05:35,835] Karthik Narayanan: So we have, like a revenue number, a margin number.

[00:05:38,686] Karthik Narayanan: We have tables, we have charts, we have a number of different charts, so essentially we want to be able to understand what's powering these what's changed with the new financial information.

[00:05:49,308] Karthik Narayanan: And whether or not Codex can help us, do some of these updates ourselves. So how I go about setting up my environment is I created a new project.

[00:05:57,414] Karthik Narayanan: And for this it's our Make Work Flow, project and apologies. If my screen is small. So I made a new project, which is our Make Work Flow project.

[00:06:05,520] Karthik Narayanan: And here I point it towards a folder that I have with some relevant information. So the first thing I want to talk about is a useful pattern.

[00:06:13,402] Karthik Narayanan: That I think not enough people use, and it's sort of setting up like durable project instructions. It allows us to establish a source.

[00:06:20,893] Karthik Narayanan: Hierarchy, and like some guardrails and guidelines for the work that we're going to go through. So for this I just set up some Codex, and like I set up a Codex instructions md file.

[00:06:31,012] Karthik Narayanan: I want to point out a couple of things in this file. I'm asked, I'm laying out the workflow. That Codex should follow, and asking it to inspect the workbooks.

[00:06:39,844] Karthik Narayanan: That I'm providing I'm asking it to preserve the original information that exists and not overwrite. The original files.

[00:06:46,553] Karthik Narayanan: I'm asking it to preserve the prior decks and visual language. Editable text tables and charts.

[00:06:51,864] Karthik Narayanan: This is really helpful because it allows Codex to learn what is in the initial PowerPoint, and allows us to preserve that sort of structure.

[00:06:59,690] Karthik Narayanan: We also want to like lay that any sort of reconciliation or changes are logged, so we know what the prior value is.

[00:07:06,119] Karthik Narayanan: Versus the future value and we also laid out some checks that we want to see before we go out and execute anything so these.

[00:07:13,248] Kenna Valdez: What, let me zoom in a little bit more. I'll also call out that we're going to be happy to share a copy of the source file or version of it. Following our session, so you can take it and adapt it with your teams. But a little bit more zoom in is the feedback. I'm getting from the the audience.

[00:07:30,131] Karthik Narayanan: Amazing, I just copy pasted the the file. I don't want to open up Xcode, but here's the zoomed in version. So it lays out. We want to inspect the workbooks and the supporting docs.

[00:07:40,422] Karthik Narayanan: And to like return an update plan rather than going out. And just building stuff for us, and we've also laid out like specific checks that we want to see.

[00:07:49,226] Karthik Narayanan: These are things that, I think are just best practices for finance, but obviously, for your use cases.

[00:07:55,058] Karthik Narayanan: You might want to iterate with Codex to build the best possible project instructions. And then feed those instructions back into your project. File.

[00:08:03,519] Karthik Narayanan: So essentially we want the updated financial model to serve as the source of truth. For both our forecast and our like DCF model.

[00:08:10,894] Karthik Narayanan: And that the actual results and forecast comparison files should be the governance of monthly performance. Discussion like that we output.

[00:08:18,783] Karthik Narayanan: And that the prior deck and baseline models are just establishing what we previously communicated.

[00:08:24,386] Karthik Narayanan: So we talked about the things that we want to highlight in our project instructions so we can go ahead and say that.

[00:08:31,018] Karthik Narayanan: We want to understand what we're working with first, so we we laid out how we want Codex to work through this.

[00:08:37,307] Karthik Narayanan: But let's take a look at the financial model that's underlying this project. So this is the updated model that's going to serve as our source of truth.

[00:08:45,940] Karthik Narayanan: For the presentation, all the numbers will be driven from here, and we have three main tabs. That, I think, are useful.

[00:08:52,743] Karthik Narayanan: We have this dashboard tab that has the output metrics that are from like you know our assumptions or DCF and whatnot. And this is our at a glance.

[00:09:01,147] Karthik Narayanan: These are our high level metrics that we want to care about. It's our executive readout. We have the update bridge that says hey.

[00:09:08,522] Karthik Narayanan: These are the numbers that are from the prior month's close. These are the numbers that are generated from the new financial data. And this is what's changed in their impact.

[00:09:18,470] Karthik Narayanan: And finally, we have the checks window, which just essentially is how you make sure all of your numbers. Tie out through your original and your ending deck.

[00:09:27,389] Karthik Narayanan: So we have, like the scenario view on our dashboard. And we have what's changed from the prior baseline on our update bridge.

[00:09:34,535] Karthik Narayanan: And we also have the fidelity check, so that gets us into our workflow. All of these files live inside of this project folder that we've pointed it towards.

[00:09:43,454] Karthik Narayanan: So now we can freely reference them inside of our Make Work Flow project. So the first thing I'm going to do is I'm going to prompt down. I'm going to drop in a quick prompt.

[00:09:53,401] Karthik Narayanan: That says, don't change any files yet. Just compare the prior deck with the may close information. And the updated financial model. Then give me a concise slide by slide.

[00:10:03,120] Karthik Narayanan: Update plan for each affected slide shows the metric or message that changes. The prior value the updated values.

[00:10:09,581] Karthik Narayanan: The ranges that we can source it back to, and inconsistency or judgment calls. We're running on fifty six. So medium, and we expect the tax like this to take a minute or two.

[00:10:19,528] Karthik Narayanan: I don't want you guys to have to sit here and watch this load. So, instead of preloaded it, what we see is, or this took a few extra minutes.

[00:10:27,589] Karthik Narayanan: I think we were running out of higher, higher model, model fidelity here, but essentially what happened is we set. We see that no files were changed.

[00:10:36,108] Karthik Narayanan: Rather, we have identified the largest issues on each slide identified.

[00:10:40,167] Karthik Narayanan: The things that existed in the prior version versus the future updated version, and the sources where we can cite them functionally.

[00:10:47,714] Karthik Narayanan: What we're doing now is, we are inspecting the changes we're asking Codex to inspect our prior deck. The baseline model and the updated model.

[00:10:55,832] Karthik Narayanan: And all the other files that we've given us, and showed us.

[00:10:59,205] Karthik Narayanan: These like highest level findings that have inspected the prior version showed us the source of the new valid information. And laid out the inconsistency in judgment.

[00:11:08,695] Karthik Narayanan: I think this is really important when you're working with finance teams. Because sometimes dropping in all these files into chat or Codex like a bit of a black box.

[00:11:18,071] Karthik Narayanan: Here we give, give ourselves the really clear visibility of the things that are wrong. Where they're wrong, how we want to update them.

[00:11:25,789] Karthik Narayanan: And where to source that information and how you can message the changes that need to happen?

[00:11:31,106] Karthik Narayanan: It shows us the facts that are laid out, the assumptions that are made, and the non reconciling numbers. So this is the most important part for a finance user.

[00:11:40,197] Karthik Narayanan: It's understanding where the numbers don't tie out, and how we can go back and source where those those different differences exist.

[00:11:47,743] Karthik Narayanan: This is sort of the level of the work of like a finance manager. So we did the finance analyst work. Now, this is the finance manager work.

[00:11:55,690] Karthik Narayanan: And finally, we're going to do the the senior level like finance managing director work as well, so we've looked at the source choices we've looked at the results.

[00:12:05,009] Karthik Narayanan: And it shows us where we find the the information. That allows us to go and do sort of like a planning stage.

[00:12:11,240] Karthik Narayanan: So we want to clarify how to handle the discrepancies that we've laid out. Here. We found a lot of areas where the numbers don't necessarily tie out.

[00:12:19,759] Karthik Narayanan: But now we want to build that sort of plan. So I'm going to drop in. An additional prompt that says.

[00:12:25,476] Karthik Narayanan: That allows us to say, using the formula driven model, which, when we specify formula, driven model.

[00:12:31,193] Karthik Narayanan: We're going to get outputs that have fully built out financial model formulas rather than just like hard coded cells. So using the formula driven model is the authority for the DCF.

[00:12:41,541] Karthik Narayanan: Keep the numbers pasted, and revise the slide by slide. Update plan if needed. So much like the last version.

[00:12:47,773] Karthik Narayanan: I don't want you guys to have to wait as this loads and chat to go through. And iterate we're going to preload some of this. Work.

[00:12:55,205] Karthik Narayanan: So we've clarified how we want the model to handle these discrepancies. We want to identify them, they've identified them.

[00:13:02,180] Karthik Narayanan: So now we want to make those changes on a slide by slide basis. So here we see that Codex is completed. The update and it reflects refreshed all eight slides.

[00:13:11,213] Karthik Narayanan: While preserving the original structure and the editable objects. So let's take a look at what the outputs look like.

[00:13:17,902] Karthik Narayanan: This is our updated deck that has all of the changes, so if we go back and we make the comparisons from the initial version oops. The initial version of our deck.

[00:13:27,163] Karthik Narayanan: We see that the charts have been updated with the relevant information. We have the new ar growth, we have the churn.

[00:13:33,852] Karthik Narayanan: The contraction we see that each of our changes are made on a slide by slide basis, and we can see that all eight slides were updated, including the version label slide.

[00:13:43,514] Karthik Narayanan: The original slide structure, theme, and objects were preserved. Every headline and ties back to the workbook. And the source range.

[00:13:51,061] Karthik Narayanan: And that the prior valuation discrepancy remains explicitly flagged. Readiness is screen-grade. We have discussion, readiness, and we have the calculation integrity.

[00:14:00,494] Karthik Narayanan: So the important part is that we didn't necessarily just generate a prettier presentation. We updated the narrative that backs the presentation.

[00:14:08,727] Karthik Narayanan: We preserve the source trail we surfaced. The inconsistency and we retained the human review grade. So.

[00:14:14,615] Karthik Narayanan: I think, when we look at the executive output, we can do a bit of a slide by slide flow through. But I don't know how interesting this will be.

[00:14:22,791] Karthik Narayanan: For the audience, rather, I think it's important to understand how we can show that level of rigor that's gone into the analysis.

[00:14:30,166] Karthik Narayanan: So we have the updated slides, and they look right and they feel directionally correct. But we also do is we generate this change.

[00:14:37,598] Karthik Narayanan: Log, and the change log shows us every single slide. The metric that has been generated, the prior number, the update. The delta and the source file and the exact ranges.

[00:14:47,317] Karthik Narayanan: This has been the area where there's been the most request. From our users, where they say hey, I know we've gone from black box, we've outlined it.

[00:14:55,778] Karthik Narayanan: But I want to be able to go through and do the same. Ticking and tying exercise that I've done my entire career, where I hold up two different presentations.

[00:15:04,754] Karthik Narayanan: And I go and see every single change I circle them, and I tie them out. It's an old banker term.

[00:15:10,242] Karthik Narayanan: But now we've we've essentially taken that entire process, and we've put it into a change log that we can quickly flip between. A excel sheet, and a PowerPoint to make sure.

[00:15:20,133] Karthik Narayanan: All of those things have actually reflected, and we're freeing up bandwidth for the teams and we're giving them a usable workflow.

[00:15:27,565] Karthik Narayanan: When they don't necessarily have the governance approvals to connect their data systems in place. Yet so. We've understood the the checks.

[00:15:35,455] Karthik Narayanan: The open decisions and we've also shown a little bit more of the traceability through our change. Logs. I think we have like a little bit of time for q and a.

[00:15:44,488] Karthik Narayanan: I'm happy to walk through the source. Hierarchy, some of the guardrails that we've built we can take a look at the source packet.

[00:15:51,863] Karthik Narayanan: We can talk about how we can update the presentation live, or how we can reconcile. And like update visual elements of the PowerPoint. But happy to spend the time.

[00:16:01,182] Karthik Narayanan: However you guys see fit.

[00:16:03,361] Kenna Valdez: Love it, thanks so much, Karthik. We have a couple of questions, and then, I think we'll have time. So continue to let us know.

[00:16:10,604] Kenna Valdez: In the chat what you would most like to see of some of those couple of options that Karthik just mentioned, or what questions you have about what you've seen so far.

[00:16:20,014] Kenna Valdez: One thing to clarify. That's come up for a couple of folks. Is that even though it says ChatGPT you're working with ChatGPT Work? In the Codex app. Could you explain the the nuance?

[00:16:30,336] Kenna Valdez: A little bit, and where, in the app? You know whether you're working in work or Codex.

[00:16:35,731] Karthik Narayanan: Yeah yeah, that's perfect. So there are two main modes to work within the desktop application. There's ChatGPT, and there's Codex.

[00:16:43,422] Karthik Narayanan: So how I like to frame this is that ChatGPT is where you can do a lot of your. Conversational back and forth, and also how you can access work, mode and Codex mode right here.

[00:16:53,775] Karthik Narayanan: So you can select between ChatGPT or Codex? Codex is a more developer or technical facing surface area. It's where you'll see code diffs, it's where you'll allow yourself to like.

[00:17:04,365] Karthik Narayanan: Go in and inspect code, and like. Look at actual here, I mean oops. I don't have it enabled here, but it'll you to have a more technical facing view of the world.

[00:17:13,950] Karthik Narayanan: So that's how you navigate between Codex and ChatGPT. And when we create a new chat, we can toggle between chat mode, which is for essentially like you want to be.

[00:17:23,593] Karthik Narayanan: Either in chat or in work, mode depending on the shape of your problem that you're facing.

[00:17:28,918] Karthik Narayanan: If you want to discuss and understand through a more text based interface like a problem that you're facing or like rewriting an email.

[00:17:36,904] Karthik Narayanan: I find chat looks like a better solution for me. But what I want to do this sort of like custom. Project hey refer back to a source file.

[00:17:45,010] Karthik Narayanan: On my desktop that has ten different documents, and I also want to use some plugins. And I want to make some tool calls. I find work to be the better surface area for that.

[00:17:55,185] Karthik Narayanan: So to zoom out you can toggle between ChatGPT and Codex. On this left side here.

[00:17:59,918] Karthik Narayanan: But when you're inside a ChatGPT mode you can select between chat or work depending on the shape of the problem that you're facing.

[00:18:07,669] Karthik Narayanan: I find, for most knowledge workers and people who don't write code for a living that ChatGPT and is the best surface area and toggling between chat and work.

[00:18:16,957] Karthik Narayanan: Gives you the right tool for the job. In front. Of you.

[00:18:20,921] Kenna Valdez: Yeah, and just to clarify we for what? You saw today Karthik used work, so you can. You can think of it as a lot of the exact same powers as Codex.

[00:18:30,043] Kenna Valdez: But for folks like me, who are not developers, and don't necessarily need to see the code so hopefully, that also addresses another question.

[00:18:38,793] Kenna Valdez: That came in about how this might look different for someone who only has access to code via via c code. So what's beautiful about work is that?

[00:18:47,729] Kenna Valdez: To achieve this, we didn't actually have to use any code directly or coding to update the presentation. Although that that may have been happening on the back end.

[00:18:57,844] Kenna Valdez: But the other question that came up is that folks would love to see a little bit more about how you go about updating the visual elements in the presentation.

[00:19:07,648] Kenna Valdez: And this was really popular in our last. Make Work Flow, too. Just how to different ways that you can share. That feedback.

[00:19:16,291] Karthik Narayanan: Yeah, I think so. I I do a lot of work with presentation design for our customers. And the area that we have a lot of success are. If you just ask.

[00:19:25,196] Karthik Narayanan: Work to go out and build you a pretty chart. You're not going to get a really strong answer.

[00:19:30,769] Karthik Narayanan: The best way to generate valuable output is to give chat an understanding of what you're looking for. You can either do this by writing out your your hopes, or your expectations.

[00:19:41,552] Karthik Narayanan: Another thing, I like to use as a feature within, within the application called app shots.

[00:19:47,004] Karthik Narayanan: So if I know, I really like some charts that we've used like some some people on my team have built.

[00:19:53,062] Karthik Narayanan: In the past, and I want ChatGPT to use those as like a framework to build their new charts off of.

[00:19:58,999] Karthik Narayanan: I can do something called, taking an app shot that takes a picture of my screen, and serves that as context. And say using this. Design a new EBITDA bridge.

[00:20:08,449] Karthik Narayanan: And when you give the model context of what you're looking for, and you tie it back to the data that is being served, you can generate a better output.

[00:20:17,596] Karthik Narayanan: So it's really like when you're teaching a new analyst or you're teaching a new associate how to build a chart. You're not going to go and say hey?

[00:20:26,501] Karthik Narayanan: Just build the best chart you can. You're going to say, here are examples of charts that we built in the past. This is where you can find that data.

[00:20:35,467] Karthik Narayanan: And if you prompt the the model to do that, it can essentially build it in a way that's better. And probably better aligned to your expectations.

[00:20:46,323] Kenna Valdez: Love it, thank you. And from our previous session we even shared a method of as you kind of develop and share. What stylistically you you like?

[00:20:55,455] Kenna Valdez: Or if there's any particular company branding you want to use. And like want to have a file that consistently kind of carries those preferences.

[00:21:04,651] Kenna Valdez: Peter diamond actually shared just having that md file.

[00:21:08,164] Kenna Valdez: So when you start a project or continue a project with Codex, you can kind of keep some of those guidelines that are probably consistent across a lot of different presentations.

[00:21:19,468] Kenna Valdez: You might do or assets that you might create. Ready to go? Another great question that came up from jenny.

[00:21:26,237] Kenna Valdez: Is she recently used ChatGPT and Codex to prepare charts for a presentation chat assess? The data and Codex help structure, the charts four.

[00:21:35,178] Kenna Valdez: Really highly regulated environments where things like GDPR and data governments are major concerns. How. Do you think about evaluating using Codex versus work for data analysis?

[00:21:46,545] Kenna Valdez: Are there any specific advantages in terms of file control data handling traceability or repeatability.

[00:21:53,731] Karthik Narayanan: Yeah, so data handling is extremely important. And we have to defer back to your organization's guidelines and approval processes around what data is eligible to be used.

[00:22:03,970] Karthik Narayanan: Within a generative AI solutions.

[00:22:05,958] Karthik Narayanan: So while Codex has a HIPAA certifications and can be flipped over into a regulated enterprise, you have to work with your admins to make sure that you have had that configuration done so.

[00:22:17,220] Karthik Narayanan: When in doubt, the best approach is to leave it out. So in regulated enterprises there are two real approaches.

[00:22:23,906] Karthik Narayanan: One you're in an unregulated workspace, in which you should keep like GDPR, sensitive or HIPAA sensitive or csi.

[00:22:30,651] Karthik Narayanan: Out of these sort of platforms, or you're in a regulated enterprise in which you should confirm with your admins.

[00:22:37,457] Karthik Narayanan: And then make sure that you have a clear guideline that you're following around. Which information that you can use within Codex or chat.

[00:22:45,709] Karthik Narayanan: Ultimately, Codex, in your experience, I'm sure you're aware of, like is a great partner to utilize for for data, data, analysis, data, science, or any sort of output.

[00:22:55,767] Karthik Narayanan: But it's important, important, and imperative, that you follow your organization's data safety processes.

[00:23:03,083] Kenna Valdez: Yeah and and maybe related to this other question is, if that you have any tips on, if you have just something more like a PowerPoint template or PowerPoint style file template.

[00:23:15,748] Kenna Valdez: How to instruct use of that asset to build presentations? So maybe we can distinguish them when you're working locally on the desktop app versus folks that have work in.

[00:23:27,841] Kenna Valdez: Cloud and web.

[00:23:29,131] Karthik Narayanan: Yeah, so if you have a something that a lot of companies have is a like a presentation guideline, so it's like a big sixty or seventy slide PowerPoint with your brand guidelines around.

[00:23:39,302] Karthik Narayanan: What presentations should look like this is gold for for for the models. When they have an understanding of what?

[00:23:45,514] Karthik Narayanan: Every slide should look like they can tie it back to the expected outputs.

[00:23:49,582] Karthik Narayanan: So if you have that that lives locally on your desktop, you can always drop that into your project file and point. Codex to that, and say, hey, this is how I want my slides to look.

[00:23:59,533] Karthik Narayanan: But the next step up is, you could write your own skill. Right that allows you to build slides in the future. That align to blossom. Companies PowerPoint template.

[00:24:08,494] Karthik Narayanan: So how you go about doing? That is, we have a preloaded skill here, which is called presentations.

[00:24:13,881] Karthik Narayanan: And in presentations we have, like a number of presentations that are preconfigured. But what you can do is you can build your own presentation skill.

[00:24:22,128] Karthik Narayanan: I don't know if I have it here. I have my own presentation skill that I built, called the OpenAI presentation studio that follows our brand guidelines.

[00:24:30,429] Karthik Narayanan: And then now, every time I want to build a presentation, I've already fed it. The understanding of what? That slide deck looks like and I asked.

[00:24:38,345] Karthik Narayanan: Codex to go and build a skill that listens to, and builds new presentations that follow these guidelines. So now, every time I want to build a presentation, I just say hey.

[00:24:47,801] Karthik Narayanan: Use this skill, and take these ten bullet points, and make me a presentation. I think that is like the best in class approach.

[00:24:54,728] Karthik Narayanan: In my experience to making a really compelling powerpoints that fit your brand. Guidelines.

[00:25:00,403] Kenna Valdez: Yeah, and if it's well, if you've especially if you've been to two or three of our webinars today, or any of the community webinars.

[00:25:08,397] Kenna Valdez: In the past couple of weeks I absolutely leverage a skill to to create.

[00:25:12,696] Kenna Valdez: With our consistently lovely orange herberty, champion style textures, all of the Make Work Flow and activator labs webinars.

[00:25:20,266] Kenna Valdez: That you, you see, in the champion community on academy. We have just a couple of minutes left.

[00:25:26,019] Kenna Valdez: So, in the interest of keeping a couple of minutes for a shameless plug for the community, I'm going to go ahead and share my screen. Really quickly here.

[00:25:35,344] Kenna Valdez: If you don't mind closing up for a moment, Karthik. Cool. Okay, let's go ahead. And. Tell you some more about that lovely community.

[00:25:43,338] Kenna Valdez: So, just as a reminder, the champion community on OpenAI academy where you probably signed up for today's session is your space to find more AI adoption trend insights.

[00:25:53,512] Kenna Valdez: Real examples just like this one and resources to help you design useful workflows that are responsibly integrated into your team systems and and lots of reusable assets.

[00:26:03,806] Kenna Valdez: If you're interested in exactly what that md file might look like or an example, you can, you can adapt and use rather than starting from scratch.

[00:26:12,648] Kenna Valdez: So sign up for future makework flow sessions. If you see more examples, that look interesting to you, or that you think you can adapt for your team.

[00:26:21,610] Kenna Valdez: We also have our insights from the OpenAI champion networks, webinars quarterly. To make sure that you are all up to speed on, the latest adoption trends that we're observing.

[00:26:32,207] Kenna Valdez: And what you can do about them as an activator. And we have activator labs.

[00:26:36,749] Kenna Valdez: If you are interested in learning a repeatable method that you can take and reuse on more workflows to just help, repeatably design develop and operationalize recurring work for your team.

[00:26:48,134] Kenna Valdez: And finally, as you build out those new ways of working, please do come back to the champion community and and share what you're building.

[00:26:56,491] Kenna Valdez: And of course, you should always follow your organization's information and security and privacy policies. But if you can share, we love to see what you build.

[00:27:06,119] Kenna Valdez: And the impact that it's had for your teams.