Menu

Real examples

How Rish Saini uses AI in restaurants, bars and cafés

See how AI supports real hospitality work, from kitchen documentation to launch campaigns. Open an example to explore the method and lessons.

Operations and teams

Pre-opening and fit-out

From site walk to organised snag list

Dictated on site walks; AI removes duplicates, sorts the items by zone and reissues the list after every walk.

A new Underdoggs venue: Underdoggs, the sports-bar brand Rish founded

How it worked

In use Used in a real business or project, with a person responsible for the result.

The problem
Finishing a fit-out means a long list of small defects, noticed while walking the site.
What went in
Rish’s dictation on each site walk.
What AI did
Removed duplicates, sorted the items by zone and reissued a branded document after every walk.
Where it stands
Used on site for a new Underdoggs venue and for Ālaya Café & Bar.
What you can take from it
Talk, don’t type. Dictate on the walk and let AI do the sorting.
Operations and owner admin

Minutes and action lists with a named owner

Minutes with a named owner for every action, and site notebooks turned into sectioned action lists.

Underdoggs: the sports-bar brand Rish founded

How it worked

In use Used in a real business or project, with a person responsible for the result.

The problem
Leadership meetings and site visits produce actions that get lost.
What went in
Meeting notes and photos of Rish’s site notebook.
What AI did
Wrote minutes with a named owner for every action and the next agenda, and turned a site notebook into a sectioned action list for the corporate team.
Where it stands
Used for leadership meetings and a new venue.
What you can take from it
Photograph, don’t retype. Notebooks, quotes and drawings go in as photos.
Staff training and onboarding

Training a new team before opening

A ten-day programme, study guides and quizzes for a new team. One study guide contradicted itself on what was vegetarian: the kind of error a person has to catch.

Ālaya Café & Bar, Shillong: a Breakthrough Design Studio project, taken from concept to opening

How it worked

Completed Finished. How widely it’s used isn’t confirmed here.

The problem
A team had to learn the menu and service standards before the first guest arrived.
What went in
The approved menu, recipes and service standards, and Rish’s pre-shift briefing notes.
What AI did
Drafted a ten-day programme, study guides, staff-friendly quizzes with answer keys, a briefing handout and the manager’s roll-out notes.
What a person checked
One study guide said every breakfast item was vegetarian while tagging eggs Benedict as non-vegetarian: the kind of contradiction a person has to catch.
Where it stands
Completed. How it was used day to day isn’t recorded here.
What you can take from it
AI writes a confident study guide. Check every dietary and allergen statement against the approved menu before a team learns it.

Menus and commercial decisions

Sales data and commercial decisions

Using sales data to inform opening hours

Six months of till data, studied in approved stages, informed decisions on service hours.

Mrs Kaur’s: his family’s bakery and café brand

How it worked

Completed Finished. How widely it’s used isn’t confirmed here.

The problem
Knowing which items, hours and occasions actually earn before changing the menu or the opening hours.
What went in
Six months of point-of-sale data, hourly sales and a table-occupancy model.
What AI did
Built a menu performance and growth study, and tested service hours against hourly sales and occupancy.
What a person checked
The study ran in stages, each approved before the next, with modelled numbers labelled as modelled.
Where it stands
Completed. The findings informed decisions on service hours and which parts of the day to trade.
What you can take from it
Stage a big analysis. Approve each step and keep modelled numbers apart from real ones.
Recipes and kitchen documentation

Recipe cards from handwriting, with a reusable Skill

Handwritten recipes transcribed onto coded cards with a reusable Skill; cards the chef hasn’t confirmed are marked as proposed.

Ālaya Café & Bar, Shillong: a Breakthrough Design Studio project, taken from concept to opening

How it worked

Completed Finished. How widely it’s used isn’t confirmed here.

The problem
A new café-bar needed recipe cards the team could follow, starting from chefs’ handwritten sheets.
What went in
Photos of the handwritten recipes, a card template with fixed fields and dish photos.
What AI did
Transcribed the recipes, merged them onto coded one-page cards, and turned the card format into a reusable Claude Skill so every new recipe comes out the same way.
What a person checked
A gap analysis marked every card the chef hadn’t confirmed as proposed. No card had an allergen field.
Where it stands
A reusable recipe-card Skill exists. Part of the card set is confirmed by the chef; the rest is marked as proposed.
What you can take from it
A neat card isn’t a correct card. Every number needs a source, and allergens need their own field.

Use the Recipe Card Starter

Menus and costing

Auditing a costing workbook

A line-by-line check of a costing workbook found price mismatches, unit errors, perfect-yield assumptions and missing costs shown as zero.

Underdoggs: the sports-bar brand Rish founded

How it worked

Completed Finished. How widely it’s used isn’t confirmed here.

The problem
A costing workbook can look complete while hiding errors that make dishes seem cheaper than they are.
What AI did
Checked the workbook line by line and listed price mismatches, unit errors, 100% yield assumptions and formulas that showed missing costs as zero.
Where it stands
Completed and written up as a deck.
What you can take from it
Ask AI to look for what’s missing as well as what’s there: zero costs, perfect yields and mixed units.

Marketing and guest communication

Marketing and content

From generic brief to venue-specific launch plan

A generic AI brief rebuilt for the real venue, city and dates, then used for the venue’s Instagram launch.

Ālaya Café & Bar, Shillong: a Breakthrough Design Studio project, taken from concept to opening

How it worked

In use Used in a real business or project, with a person responsible for the result.

The problem
AI marketing output often fits any venue in any city.
What went in
A generic brief from one AI tool, plus the real venue, city, guests and launch dates.
What AI did
Claude critiqued the generic brief and rebuilt it, then produced the launch calendar, content pillars, reel hooks and captions in the brand’s voice.
Where it stands
Used for the venue’s Instagram launch.
What you can take from it
Push back on generic. Give the real venue, city, dates and guest, and reject anything that could be about anywhere.
Marketing and guest retention

A WhatsApp anniversary campaign

AI built and sent the bakery’s twentieth-anniversary WhatsApp campaign inside its marketing platform, within Meta’s daily cap.

Mrs Kaur’s: his family’s bakery and café brand

How it worked

Completed Finished. How widely it’s used isn’t confirmed here.

The problem
The bakery’s twentieth anniversary needed a WhatsApp campaign to its guest list, within Meta’s daily sending cap.
What AI did
Built the campaign and sent it inside the bakery’s marketing platform, working around the daily cap.
Where it stands
Sent.
What you can take from it
AI can do the clicking in the tools you already pay for, with your approval.
Brand foundations and writing rules

Brand audit and writing rules for a 20-year-old bakery

AI logged the brand’s facts and contradictions and helped write the rules that later captions, replies and menus follow.

Mrs Kaur’s: his family’s bakery and café brand

How it worked

Completed Finished. How widely it’s used isn’t confirmed here.

The problem
Facts about the brand had drifted across old files, menus, table cards and online listings.
What went in
Every old brand file, menu, tent card and listing.
What AI did
Logged the facts and contradictions, reorganised the brand folder into one master set and helped write the rules that later captions, replies and menus follow. It then checked the website against them and listed fixes.
Where it stands
The writing rules are approved and inform the review-reply drafts. The website fixes are a separate step.
What you can take from it
Write the rules once. A brand bible and saved preferences mean every later output starts on-brand.

Use the Brand Voice Starter

Guest reviews

Drafting review replies in the brand’s voice

A scheduled routine drafts a reply to each new Google review under a written standard; safety and legal cases always go to a person.

Mrs Kaur’s: his family’s bakery and café brand

Details

In development Underway: being built or tested, not yet in regular use. Shown as a method, not a result.

The problem
Every review should get a considered reply in the brand’s voice, but a few need a person rather than a public answer.
What went in
The brand’s writing rules, a written reply standard (voice, what is never said in public, which reviews to hold) and each day’s new reviews on both Google listings.
What AI did
A scheduled routine drafts a reply to each new review under the standard and logs it for review.
What a person checked
Reviews that raise safety or legal issues are always held for a person. Rish reads the replies and flags anything that’s off.
Where it stands
Replies have been posted with Rish’s go-ahead. Posting them automatically is being fixed.
What you can take from it
The rating isn’t the risk; the subject is. Write down what must never be said in public and what always goes to a person.

Lessons and teaching examples

Checking lesson · Menus and kitchenOne recipe, written down twiceA chef's handwritten ciabatta sheet, photographed in April 2026, and a printed recipe card for the same bread, made in June. They disagree on almost every line, and the card even disagrees with itself on the bake.See the walkthrough

Some worked examples inside the free kits are made up for teaching and labelled that way: a fictional café’s caption in the Brand Voice Starter, and rough masala chai notes in the Recipe Card Starter.