Exploring AI in My Restaurant—Ask Me Anything!

[OSP1]

My name is Eric, and I’m the owner of Old Scratch Pizza in Dayton, OH, and a Toast Customer Advisory Board member. AI has become a daily tool in our restaurant, helping with everything from employee communication and marketing to streamlining operations and reputation management. Tools like ChatGPT and Claude assist with writing, problem-solving, and automating tasks, saving time and effort.

I also use AI to extract deeper insights from Toast data, giving us answers to questions that were previously unavailable. If you’re curious about AI in restaurant management or reporting, I’m happy to share my experience—ask me anything!

[OSP1]

We’re using AI in lots of ways to make operations and the business more efficient:

Order Fulfillment Analytics – We were having trouble getting a report out of Toast that would let us know how our restaurants compared against each other  in our order fulfillment times. We now export an Order Details report for multiple restaurants, which is a CSV file of over 8,000 lines, and upload it into GhatGPT. With a prompt we developed (let me know if you want to see an example of the prompt), ChatGPT will analyse the data and make a report and graph analyzing our fulfillment times for the prior week. We use this as part of our weekly manager review. I could have never done this without AI.

Marketing & Content Creation – AI speeds up social media, ad copy, and promotional content, keeping our messaging consistent. We have uploaded examples of our writing styles into ChatGPT and Claude, and the programs aid in the first drafts of social media content, email promotions, etc. It also helps us brainstorm ideas for content and promotions.

Survey Analysis & Customer Insights – We often do surveys of our customers and employees (often just with google forms). In the past we would have never been able to have decent analysis. Now we upload the results into AI platforms, and have amazing insights and feedback, including comparisons to industry standards.

Reputation Management - As I mentioned earlier, we have uploaded lots of examples of our writing and brand voice. We use this to quickly create responses to Google and Yelp reviews that are thoughtful and kind. 

Financial Analysis - One of the most interesting things to do is to take your financial data, like your P&L, and upload it into GhatGPT (or Claude, or Gemini, etc). Then, ask "Please perform an analysis of my financial statement. Compare my performance to other restaurants like mine. Make recommendations and actionable tasks for how I can improve problem areas. Ask me any questions necessary to improve your results". You will be amazed by the answers and recommendations! 

HR - We use it to quickly create first drafts of Job Descriptions, performance review forms, or any other HR tool that you might not have. 

Legal - I have even used it to help me negotiate a lease and to write a letter that helped get me out of a contract with a software vendor. 
I hope that helps. There are so many things. If you have a specific use case you are thinking of, let me know and I could point yo in the right direction.

[seabird]

Hey Eric, have you used AI to assist with anomaly detections (like voids, refunds, account categorization changes in the accounting software, etc.?)

[OSP1]

I have not tried this. Particularly not with voids or refunds. However, for account category changes I would think it would be pretty straightforward. We don't use Quickbooks, but I assume you could download a GL report in CSV format, and then upload it and ask ChatGPT "Rework this spreadsheet for me converting all entries of GL4509 to GL4508", or "Find all entries that mention Hardware and reclassify to GL Restaurant Repair and Maintenance", or something like that. But, whatever question you have of the data, no matter how many entries it is, GhatGPT can find you the answer. The first step is to get it into a format that the system can use. Most of the CSV downloads from Toast work great. then, just keep chatting and asking questions until you find the info you need.

[rcmck]

---

@OSP1 wrote:
I also use AI to extract deeper insights from Toast data, giving us answers to questions that were previously unavailable.

 How are you doing this specifically? Just manually downloading and importing the data? Or are you using an integration or Toast API? 

[OSP1]

Today it is a manual process. We have fixed prompts set up in a folder in ChatGPT, and we just download a Toast report and upload. Some sort of integration would be great, but it isn't there right now. Maybe some day this level of AI might be built into Toast so that we could query our data. Because you might find it interesting, here is an example of an Order Fulfillment prompt based on the Toast Order Details report:

"Analyze restaurant order fulfillment times using the provided CSV.

• Group data by 'Location', 'Check #', and 'Date'.
• Calculate the difference between the earliest 'Fired Date' and the latest 'Fulfilled Date' for each unique check.
• For each location, calculate and display:
  • Total number of checks Number and percentage of checks fulfilled in 20 minutes or less
  • Number and percentage of checks fulfilled over 20 minutes
  • Format the percentages to two decimal places with a '%' sign.
• Then, create a side-by-side bar graph comparing each location for each day of the week"

Toast can give rudimentary fulfillment data, but this gives us a specific measurement for our teams to work toward each week. I hope this helps!