Use ChatGPT Work to build custom forecasting apps

> TL;DR: OpenAI's finance team used ChatGPT and Sites to build an interactive financial forecasting app, transforming the monthly forecasting process from waiting for data and updating spreadsheets into a collaborative, dynamic site. ## Challenge: The pressure of month-start forecasting Hi everyone, I'm Jackson, and I work on the data science team at OpenAI, partnering closely with the strategic finance team. Many of you in finance teams are probably familiar with this scenario: the start of each month is very busy — waiting for data to land, updating spreadsheets, coordinating with product, marketing, and go-to-market teams on how to update the forecast, all while trying to build a solid, reliable projection to present to the CFO. OpenAI is no exception. But the team kept asking itself: How can we make this process better? ## Goal: An interactive, collaborative forecasting interface The team wanted to solve this: build an interactive interface that can be shared with finance partners, collaborate with them, and ultimately update the forecast. The first thing Jackson did was prompt ChatGPT: > Hey, let's kick off this month's forecast. Pull the latest actuals, update our data science baseline, and use Sites to build an interactive financial forecast. Sites is a tool frequently used inside OpenAI — essentially a way for anyone to build a website no-code and host it without needing any infrastructure. ## The generated forecasting site What ChatGPT generated includes: - Data science baseline - Prior forecast - Different scenarios you can build - All of this shown at a summary level There are also corresponding visualizations, so Jackson can quickly scan trends and see how the forecast is tracking for the second half of the year. ## Adjusting the forecast by account The site supports viewing and adjusting account by account. Using Apex (one of the larger accounts) as an example: - You can see the baseline data - You can see a green line - You can adjust that green line directly in the interface The key point is that even though you're adjusting the forecast for this account, the whole model updates in the background, and the summary numbers adjust accordingly. ## Value and impact This approach saves the finance team a lot of time, allowing them to do more strategic forward-looking analysis or initiatives, rather than being stuck in tedious spreadsheet updates and waiting on data. ## Source Use ChatGPT Work to build custom forecasting apps - YouTube (https://www.youtube.com/watch?v=mOAHWk24D28)