A useful chat wrapper around DoorDash is feasible, but a fully integrated one is constrained by DoorDash's access model.
You could ship a useful v1 without deep DoorDash integration. The easy version is a chat UI plus your own memory layer: save meals, tag cuisines, record price ranges, and let users say things like reorder the spicy chicken bowl, find me something like that but cheaper, or what did I like from Thai places. That part is ordinary app work.
The harder part is doing this as a true DoorDash wrapper that can reliably read order history, compare live menus, and place orders end-to-end on behalf of the user. The public evidence points to DoorDash exposing official APIs mainly for merchants and delivery/logistics partners, not for consumer-side assistant workflows.
Create Menu, Update Menu, Confirm Order, Cancel Order, Get Store Info, and Get Store Menu Details at https://developer.doordash.com/en-US/api/marketplace/. It also says: Our Marketplace integration pipeline is currently at capacity. We are not accepting new partners at the moment while we develop self-serve tooling for merchant onboarding. That is a strong signal that even partner access is controlled, not a simple open consumer API.https://developer.doordash.com/en-US/docs/drive/overview/about_drive/ says: Production access to the Drive API is currently restricted, and we cannot provide a timeline for certification following development. The tutorial at https://developer.doordash.com/docs/drive/tutorials/get_started/ also references credentials like developer_id, key_id, and signing_secret, plus a request for production access. That is normal B2B API onboarding, not something a chat wrapper can casually plug into.https://about.doordash.com/en-us/news/ai-powered-merchant-tools, DoorDash says its new self-serve onboarding helps merchants launch more than 35% faster and that AI-powered websites are seeing order conversion rates of nearly 10% on average. Useful signal: DoorDash is absolutely investing in AI, but the public tooling it is announcing is for merchants, menu content, onboarding, and direct ordering channels, not a consumer memory/recommendation assistant API.https://play.google.com/store/apps/details?id=com.dd.doordash&hl=en-US shows 4.7 star, 5.89M reviews, 50M+ Downloads, more than 310,000 menus, 55,000+ grocery, convenience & retail stores, and 4,000+ cities. That scale is useful for discovery, but it also means menu coverage, price changes, availability, substitutions, fees, and store-specific options are highly dynamic. A chat product can reason over remembered meals and current search results, but production-grade ordering automation has a lot of moving parts.https://www.doordash.com/robots.txt from the shell returned a Cloudflare challenge page with visible text Just a moment... and Enable JavaScript and cookies to continue instead of a plain robots file. I also could not cleanly retrieve the consumer terms content as plain HTML from the public support URL because the page shell only exposed Loading and Sorry to interrupt. That does not prove a legal prohibition by itself, but it is strong operational evidence that headless/scripted consumer scraping will be fragile.good late-night, and rough price bands. Use live search only where you can do it compliantly. This does not need deep DoorDash access to be valuable.save this meal, pasted receipts, screenshots, forwarded order emails, or a browser extension that extracts item names only when the user is active.like last time but cheaper, closest thing to the spicy chicken bowl under $18, what have I liked that travels well.Not very hard to build a useful version. Harder to build a truly integrated, reliable DoorDash wrapper. The public evidence supports a product strategy where you own the memory/recommendation layer and keep DoorDash integration thin or optional, instead of betting the product on unofficial consumer automation.