Craftable Intelligence Operator AI

Know what's slipping before the shift ends.

Operator AI ranks tonight's biggest revenue leaks and opportunities by dollar impact, tells your team exactly what to do, and measures whether it worked. Built on the Craftable data you already have.

The Operator AI Insights screen for Main Street Diner: total current opportunity of $62,920, split across Sales $25,850, Labor $12,700, Food & Beverage $19,680, and Controllables $4,690, with the top Sales insights ranked by potential impact — lunch sales down 8% worth $12,400 a month, breakfast up 12% worth $6,500, a worsening beer attach rate worth $4,200, and dessert attach climbing worth $2,750.
The reframe

Your reports tell you what already cost you money.

Dashboards show last week. Month-end reconciles the damage after it's done. Between those two, a server's void rate creeps, a vendor's price jumps, an item gets over-portioned — and the margin's gone before anyone reads a report.

The month-end loop

You find out after the money's gone.

The leak runs all month. You see it in the P&L weeks later, once it's already spent.

  • 01 Shift happens
  • 02 Data lands in a dashboard
  • 03 Month-end reconciliation
  • 04 You spot the leak — too late to catch it
The same-shift loop

Operator AI works before the next shift.

It ranks tonight's problems and opportunities by dollar impact and hands your team a move they can make now.

  • 01 Runs nightly on your data
  • 02 Ranks every leak and opportunity by dollars
  • 03 Tells your team what to do
  • 04 Measures whether the action worked
How it works

From your data to tonight's punch list.

01

Runs nightly on the data you already have.

POS, labor, invoices, inventory — no new integration, no new work for your team. It's live from day one.

02

Ranks every opportunity and leak by dollar impact.

You get a short, prioritized list, not a wall of charts. Each card shows the number, the reason, and the suggested action.

03

Measures whether the action worked.

Every insight tracks views, engagement, the action taken, and the revenue recovered — so you can see what actually moved the number.

Every card cites its math. The numbers are computed deterministically and fully auditable — the AI ranks and prioritizes, it doesn't invent numbers. No black box.
The insight detail

One card. One number. One move.

Open any insight and Operator AI shows the summary, the evidence, and exactly how it got the number. This is the real beer attach rate insight: attach fell from 22% to 18% on dinner tickets, worth about $4,200/mo if not recovered — with the math laid out so your team, and your CFO, can check it.

Operator AI's insight detail for a worsening beer attach rate: a summary explaining attach fell from 22% to 18% over five days on nights newer servers carry the dinner shift; a potential impact of $4,200 per month; an evidence chart showing the 22-to-18% drop since detection; and a 'How we got there' breakdown — a 22% baseline across 380 dinner tickets a week at a $7.50 average beer ticket is about $2,510 a week, versus $2,052 a week at 18%, a $458 weekly gap or roughly $4,200 a month if not recovered — plus a box to ask follow-up questions and buttons to dismiss or mark the insight as actioned.

Every insight is quantified, benchmarked against your own floor, and paired with a move your team can make before the next shift.

New in Operator AI

The first F&B system that proves the fix paid off.

Most analytics stop at the insight. Operator AI keeps going. Like an ad platform tracks a click to a conversion, it tracks each insight from view → open → action → outcome. The week after your team acts, it checks whether the metric actually moved — and if it didn't, the card comes back with a worsening badge. You stop guessing whether the fix worked. You see it.

  • GMs watch the list get shorter.
  • Area managers see which stores are acting.
  • F&B directors and CFOs get a closed-loop record of ROI — a measurement view that rolls dollars identified, actions taken, and revenue attributed into one screen.

See the closed loop on your own numbers.

Book a demo and we'll walk your data — leaks, opportunities, and what a recovered fix looks like.

Book a demo
Breadth

Eight ways it finds money you're leaving on the table.

Underperforming servers & shifts

Who's slipping versus your own floor average.

$10,000+per server, per year

Opportunities worth doubling down on

Positive trends to reinforce while they're hot.

up to $4M+single item to full portfolio

Vendor price swings

Cost-driver shifts caught at the SKU level.

$2,000–$5,000per SKU

Purchasing vs. usage waste

Buying more than your recipes actually use.

$5,000–$20,000per item

Cost leaks

Discount and comp anomalies that add up.

up to $66,000per year

Inventory shrink

Shrinkage tracked item by item.

$1,000–$4,000per item

Void spikes

Void rates running above your floor.

up to $69,000per year

Labor ROI & overtime drag

Where labor spend isn't returning.

up to $91,000per year

Across the April 2026 pilot, Operator AI surfaced over 5,000 quantified insights — 31 insight categories running nightly against real operator data.

The finance unlock

Numbers your CFO can defend.

Auditable math.

Every insight cites the data behind it. The numbers are computed deterministically — the AI ranks and prioritizes, it doesn't invent them.

Benchmarked against your own floor.

Every comparison is within a single store, against that store's own daypart and revenue-center averages. Never against other brands. Never industry data. Your numbers stay yours.

Your data stays isolated.

Insights are stored per store, isolated by tenant. Your data is never used to train models for anyone else.

Multi-unit & hotel

Every store. Every outlet. Every night.

Whether you run 12 restaurants or a hotel with 6 outlets, Operator AI treats each location as its own store — ranking each one's insights and, with action attribution, showing you which locations are acting on them. Triage the whole portfolio from one feed.

The Craftable advantage

It already knows your business.

Operator AI isn't a bolt-on that needs a data project to start. It runs on the POS, labor, invoice, and inventory data already flowing through Craftable — the only platform built for the full hospitality back office. That's a wider data surface than a standalone analytics tool will ever see, live from day one.

One data surface, feeding Operator AI

POS
Sales, voids, comps, dayparts
Labor
Hours, overtime, schedules
Invoices
Vendor pricing at the SKU
Inventory
Usage, shrink, recipe cost
Ranked, benchmarked, and measured — every night
We'd have a gut feeling something was down. But proving it took a deep dive — pulling the query, building the data, making it digestible. Now we spend more time talking about what to do, rather than proving the problem exists first. — Director of Beverage, multi-unit restaurant group · Operator AI pilot
Built on the Craftable platform

The back office behind operators at scale.

10,000+
operators on the platform
$5B+
in purchasing managed
60K+
hours saved in 2024
5,000+
quantified insights surfaced in the Operator AI pilot
$62,920
open opportunity surfaced in a single restaurant
$145,678
open opportunity surfaced at one hotel in a month

Trusted by operators including bartaco, Hôtel Swexan, TC Restaurant Group, Woods Coffee, and José Andrés Group.

Questions

What operators ask before they buy.

Does it make up numbers?
No. Every figure is computed deterministically with rules-based arithmetic and cites its source data. AI ranks the insights by impact and writes the plain-language action and context — it never generates the numbers themselves. Open How we got this on any card to trace the math.
Do I need a new integration?
No. It runs on the data already in Craftable — no new integration and no new work for your team.
What data does it use?
Your POS, labor, invoice, and inventory data — the same data already flowing through Craftable.
How are benchmarks calculated?
Within each store, against that store's own floor average by daypart and revenue center. Never cross-brand and never against industry data.
Is my data used to train models?
No. Insights are isolated per tenant. Your data powers your answers and never trains a model for anyone else.
How often does it update?
Nightly, for every store. You start each shift with a fresh, ranked punch list.
Do I need a data analyst to use it?
No. It delivers a ranked punch list with a dollar figure and a suggested move — not raw charts to interpret.

See what Operator AI finds in your numbers.

Book a demo and we'll walk your data.