Multiplier l'impact des équipes : Stephanie Anani, ingénieure solutions, OpenAI

[music] >> Hi everyone. I'm Stephanie, a solution engineer here at OpenAI. I work with some of our largest customers to show them how they can get value from this technology. So, Lee has spoken to you about how workspace agents can really transform your team. In this section, I want to get a bit of a deep dive into how your individuals can get lots of value from AI. You tell us the same thing over and over again. Your employees are spending time doing work that needs to happen, but work that doesn't actually require human judgment. This is where we think AI comes in. AI can free up their time so that they can work on the things that actually drive value in your business. A recent survey we did found out that 75% of ChatGPT users were able to do things that they couldn't do before. That's not just doing your job faster. That's increasing the operating level, and we think that your employees should be working at this increased capability. In order to do that, we think they need support on five key dimensions. And today, I really want to specifically share why ChatGPT is the platform for financial services. First is context. To get your jobs done, you need access to trusted sources with up-to-date information. We've worked hard to make sure that we've connected to some of the most trusted pieces of information in financial services. So, for example, we have app connectors to Dow Jones, LSEG, and S&P. This is just the start of what we're doing, and we're continuing to work on this. Additionally, we know that a lot of financial services work happens in Excel. So, we built chat GPT in Excel that you can interact with your workbooks in natural language. What is underpinning these capabilities is GPT 5.5. We went a step further this time, and we decided to embed financial services processes directly in the intelligence of the model. In order to test this over time, we developed OpenAI Banker Bench. This gives you the ability to see how frontier models perform on financial services tasks. You can see that GPT 5.5 is state-of-the-art above all the other frontier models. Now, I love telling you about all these advancements, but I actually want to show you. So, let's head over to the demo. So, I am an investment analyst at Blossom Bank. My manager has just sent me this message. We need a fresh view on QXO for an IC meeting tomorrow. So, I go on to chat GPT, and the first thing I know, if we ever want to have an effective investment decision, is gathering as much context as possible. You all know and love deep research. So, we're going to use this capability to gain more information around QXO. We're also going to have the ability to connect to some of these apps so that we ensure that our trusted sources are used in this deep research. Lastly, we can connect it to sites that we trust and we respect. So, here I have my prompt. We're going to create an investment dossier deep dive, and I I focus on things like SEC filings, earning transcripts, and investment presentations. So, what happens behind the scenes is that ChatGPT creates a plan. At this point, you have the opportunity to go in and change anything. But, I'm pretty happy with what I see here. So, let's start the report. Once the dossier is created, we can go through it and try and find key information that's necessary. So, one thing I realized is that QXO is acquiring Kodiak. That is going to have a massive difference in whether we should invest in them or not. These reports are great, but how do we actually turn this into numbers? This is where we use ChatGPT to create an Excel workbook for our investment statement. So, here we have a specific skill that we're using, the Blossom Bank's three-statement model. In this skill, we've decided exactly how we like our three-statement models done. And so, instead of me having to write a detailed prompt telling ChatGPT exactly how to do things, all of that knowledge is within the skill, and then I can write a brief prompt saying specifically what I'd like to do. ChatGPT is able to create an Excel workbook. Now, what I love about this is that it's not just a black box. We can see that ChatGPT has left comments about assumptions and sources. Another really great capability is that all of the numbers actually have formulas. So, you can go in and actually audit these capabilities yourself. So, I've had a meeting with my manager, and we've decided that we're going to take a very specific approach with this IC. We want to be able to measure the bear, base, and bull case for QXO. So, I've dumped all of that context into um ChatGPT and Excel. We see that it reads the context, it decides what needs to be done, and then I get a result. I get a summary of what changed, key numbers, and insights to glean from. And this is incredible. Before, this might have taken hours or days. Now, I'm able to do this in minutes with ChatGPT and Excel. So, now we can move almost as fast as the markets. This is great, but for an IC, we know that we really need to take out all of the important signal and present it in a way that can just inform the decisions that are going to be made. Our next step on this journey is actually transforming this into a presentation that's ready for the IC. The first thing we use is a Blossom Bank deck build skill. So, this gives us the ability to put in all the know-how as to how Blossom Bank likes to create decks. Then, I also put on extended thinking because I want ChatGPT to really take its time to make sure that we're getting the output that we want. Lastly, I make sure I add all of that context. So, the deep research report and also the valuation that we've done. ChatGPT is able to create this deck. So, let's have a look at at what's been created. Here, we can see there's a clear decision, approve with conditions. What I also love is that we've also created charts in the presentation. So, you can see clearly what the decision points are and the valuation outcomes. Another great capability is that in the presenter notes, there actually comments around why decisions were made. So, you can actually audit this process end to end. So, what we've been able to see is not just a feature run-through. It's going from trusted contacts to deep research, going from deep research to a financial model, and going from a financial model to decision points that actually move the business forward. AI does not replace the need for human judgment. It just frees up the day so that you can use the time to make those judgments. >> [applause]