Friction Audit

Find the work worth fixing before you buy the technology.

A fixed-fee operational diagnostic.

We walk your operation, map how the work actually moves, and price every friction point. Then we tell you which ones deserve technology and which ones don't.

Most AI assessments start with a technology inventory and work backwards to a use case. That order always finds work for the tool that was already chosen. This starts with the work, and the people doing it, and it's run by chefs rather than by a software vendor.

Variance isn't the defect. It's the product.

Time studies and engineered labour standards need a task that repeats often enough to be measured and held. That works, and it works well, in a kitchen running a fixed menu. Point it at a room that's a wedding on Saturday, a conference on Monday and a member dinner on Wednesday, staffed largely by people who've never worked it, and the method has nothing to hold onto. This is usually read as the operation being immature. It isn't. Someone is paying precisely because this event isn't like the last one.

When to run it

Before the money commits. A clubhouse renovation, a new outlet, a system purchase, a concept refresh: the decision gets locked early and the first guest arrives two years later. Between those two dates nothing gets tested.

We run in that gap. The audit is the See and Shape half of how the studio works, sold on its own, so you can find out what's true before you commit a capital plan to it.

It is a fixed fee and a fixed scope. Not a retainer, not software selection, and not a referral to a vendor we have a relationship with. We don't have one.

What we watch, and what we count

Watched
  • Peak and off-peak service, on the same room
  • Prep, production, and the handoff to the pass
  • Receiving, and at least one inventory cycle
  • Close and the reporting that follows it
  • Event orders, and every revision of them
  • The workarounds nobody wrote down
  • Manager interruptions during service
  • Every screen someone has to touch to do their job
Counted
  • Touches and handoffs
  • Wait states and queue time
  • Travel distance
  • Rework and error recovery
  • Duplicate data entry
  • Decision latency
  • Task fragmentation
  • Manager intervention rate
  • Labour minutes per cover or per event
  • Throughput constraints
  • Guest delay
  • Lost selling capacity

Where AI earns its place

Every friction point resolves to exactly one of six verdicts. The rubric exists to make the call defensible, not to maximise how much of your operation ends up automated. Two of the six involve no technology at all, and one of them says leave it alone.

Eliminate

The work shouldn't exist. A report nobody reads, a second entry, an approval that has never once been withheld.

Redesign

The work is real and the process around it is wrong. Automating a broken flow only makes it fail faster.

Standardise

Some variation is genuine and some is drift. Telling them apart is the judgment, and getting it backwards sands the character off your service.

Automate

Repeating, backstage, correctable, clean data. Ordinary software usually beats a model here and costs less to run.

Augment

A person stays accountable and can overrule it, and the shift still works when the tool is down.

Human preserve

Removing the friction would remove the hospitality. Not a concession, and not a stage on the way to automating it later.

What you get

  • A current-state map of how the work actually moves, per outlet
  • A friction register, each line priced in annualised dollars
  • Root cause named before anything is prescribed
  • Every opportunity sorted into one of the six verdicts
  • A ranked backlog, sequenced 30, 60, 90
  • The where AI does not belong list, in writing
  • A metric dictionary: formula, source, cadence, baseline, and what the number can't tell you
  • One named person accountable for each recommendation

The honest part

Sometimes the answer is that AI doesn't belong anywhere in the process you asked about. That happens, we'll say so in writing, and the audit is the same price when it does.

A pre-service tasting is a good example. A model can flag that a sauce is outside its normal colour range. It can't decide whether this plate is right for this room tonight, and building toward a kitchen where it does is building toward a venue where nobody tasted the food. That has been tried at scale. You can see how it went.

The audit doesn't include implementation. If the work that follows is something we build, we'll tell you what it costs before you decide. If it isn't, we'll tell you that too.

When this is the wrong call

  • Fixed-menu, single-format, high-volume service. Classical industrial engineering is the better instrument there and it's more mature. Use it.
  • Operations that haven't stabilised the basics yet. Ordering, receiving, scheduling and closing should work before this is worth running.
  • Anywhere the real goal is cutting headcount. This will route almost everything to human preserve and you'll find it obstructive.
  • Capital that's already poured. This informs what gets built, not what's already in the ground.

Demos, not memos.

Private clubs, resorts, outdoor and RV hospitality, and the teams behind private and experiential events. Fixed fee, fixed scope. Bring us the operation you already know is harder than it should be.