Dishoom's website has a file most restaurant groups haven't even heard of yet: llms.txt — a structured document that tells AI tools exactly how to answer questions about the business, rather than letting them guess.
Most hospitality brands aren't thinking about this at all. Dishoom already is.
But that same file reveals something else. It explicitly instructs AI tools not to quote "outdated, generic booking rules" site-wide — because the rules genuinely vary café to café. That's not a guess on my part. It's Dishoom's own team, in their own words, flagging that the detail changes faster than a general answer can keep up with.
Which raises the actual question worth asking: if the team already knows this varies enough to warn an AI about it, is it equally clear to a first-time visitor landing on a specific location page — before they turn up expecting to book a table the normal way?
That single observation, and three others like it, are laid out in a one-page breakdown below — built entirely from what's publicly checkable: Dishoom's own published pages, their own llms.txt file, and independent reviews. Nothing here claims any insight into covers, revenue, or internal reporting.
It also isn't all critique. The same research surfaced a genuine strength worth naming directly: a restaurant group investing this early in AI-readiness, and this deliberately in brand storytelling, rarely leaves the rest to chance by accident — which changes what the fix actually is.
Get the full one-page breakdown
All four findings, plus what I'd check first, in order — free, no further gate once downloaded.
Part of an ongoing series reviewing what's genuinely public about a brand's marketing — never internal data. If you'd like to see how your own business reads from the outside, get in touch.