Mixing Ordinal Mentions in Food Ordering Agent

Hi, just asking for some design advice. Consider a fast food agent and the following story:

  1. User: I would like a cheeseburger
  2. Bot: Okay, I’ve added cheeseburger to your order
  3. User: Can I get extra cheese, extra lettuce and no pickles on it. Also, I would like a large soft drink and large fries
  4. Okay, I’ve added extra cheese, extra lettuce and removed pickles from your cheeseburger. I’ve added a large soft drink and large fries to your order. Will there be anything else?
  5. No, that’s it.

Where the order can just be a listslot of “items” in their order.

So the issue that I am thinking about is in line 3, where the user may want to modify something in their order that existed before. However, I struggle to see how we could get ordinal mentions to meld with the other sentence, where the user just adds the drink and fries to their order. More troublesome would be if they wanted to do something like “Can I get extra cheese on my cheeseburger” and for the system to recognize that this is a cheeseburger already on the order, and not a new cheeseburger. Does anyone have any suggestions?

Also, for reference, this is for a ASR system, not a text system, so I won’t really have the option to make a bunch of small intents and forms since talking takes a real long time.