R&D Active · Helsinki Design Partners

Your kitchen runs
on assumptions.
Orbis runs on data.

Orbis is the intelligence layer independent restaurants have never had — connecting what you expect to sell, to what you prep, to what you order, to what it costs you in labour. Before the week starts.

ORBIS · SERVICE VIEW · TUE 07:45SHADOW MODE ACTIVE
COVERS FORECAST · THIS WEEK
Mon
142−8%
Tue
196+4%
Wed
168−2%
Sat
287+12%
MISA LIST · TUESDAY PREP
Salmon fillet
Last 4 Tue avg · rain forecast −6%
28–34
portions
Lentil soup
Lunch peak signal · +festival nearby
18 L
batch
Beef tenderloin
Pre-orders confirmed: 6
14–18
portions
INVENTORY CHECK · AUTO-DEDUCTED
Salmon fillet6 kg stockORDER 8 kg
Red lentils12 kg stock
Beef tenderloin2.4 kg stockORDER 5 kg
LABOR COST VIEW · TUESDAY
Forecast revenue€3,920
Shift cost (TES)€1,140
Labor as % revenue29.1%
SHADOW MODE — Orbis is watching. Not yet acting. Owner reviews all suggestions.
Why Orbis exists

Independent restaurants make the same four expensive mistakes. Every week.

Not from carelessness. From operating without the data layer that makes better decisions possible. Orbis is that layer.

01
Over-prep

You prepped for Saturday's crowd on a Tuesday that looked busy. Forty portions of salmon went in the bin at close. The data to predict Tuesday existed. Nobody was reading it.

02
Overstaffing

Five on the floor for a service that needed three. Payroll for that shift exceeded what the kitchen earned. The rota was written on a gut feeling about a day that didn't arrive.

03
Blind ordering

Supplier invoices arrive and you've overbought again. Or you ran out mid-service. Ordering without a demand forecast is a guess dressed up as a process.

04
Compliance cost

TES collective agreement rules — Sunday premiums, evening rates, three-week averaging — calculated manually every payroll cycle. Hours spent on arithmetic that should be automatic.

The intelligence loop

One chain. Runs before your kitchen opens.

Most restaurant software gives you tools to react. Orbis gives you a loop that runs before the shift starts — so decisions are made on data, not instinct.

What you sold — and why

Orbis reads your POS history alongside weather, confirmed reservations, local events, and day-of-week patterns. Not just what sold — the signal behind it.

POS integration · External signals · Reservation data
What you'll likely sell — with reasoning

A demand forecast per dish, with the logic shown. Not a black box. "Meatballs 30–50 portions — last 4 Tuesdays, rain forecast −10%, match day nearby +15%." You see why.

Demand forecast · Reasoning shown · Override logged as training data
What to prep. How much. In what order.

The misa list — quantities per dish, sequenced by station and prep time. One tap to confirm. Every correction you make feeds the next prediction.

Mise-en-place list · Station sequencing · Confirmation in one tap
What that takes from inventory

Confirmed misa quantities are deducted from live inventory using your recipe graph. Shortfalls are identified automatically — you don't count backwards from what's on the shelf.

Recipe-locked deduction · Live inventory · Shortfall flags
What to order. From which supplier. At your price.

A draft purchase cart per supplier, quantities in pack sizes, prices pulled from your own invoice history. You approve. Orbis learns your supplier relationships over time.

Draft cart · Supplier price history from invoices · Human approval
What the shift costs before it starts

Demand forecast → staffing need per hour → draft rota that is TES-legal by construction. Labor cost as a percentage of forecast revenue, visible before a single hour is worked.

TES-compliant scheduling · Labor cost view · Draft rota for approval
Shadow mode — live in R&D

Before Orbis acts, it watches.

The biggest fear any restaurant owner has with a new system is being wrong at the wrong moment. Orbis was designed to earn trust before it asks for it. Shadow mode runs the full loop silently alongside your kitchen — forecasting, calculating, comparing its predictions against what actually happens. You see the accuracy improve in real time. You decide when you trust it enough to act on it.

01ShadowOrbis runs silently. You see results vs actuals.
02SuggestOrbis surfaces suggestions. You approve each one.
03AssistOrbis acts on stable patterns. Edge cases escalate.
04AutonomousOpt-in. Low-risk ingredients ordered without touch.
SHADOW ACCURACY · WEEK 3LEARNING
Covers forecast accuracy84%
Misa quantity accuracy77%
Inventory shortfall prediction91%
Week 1: 20% · Week 2: 54% · Week 3: 84%
Each correction you make is a training event.

What's live now.
What's coming next.

Orbis is in active R&D with Helsinki design partners. This is the honest picture of where the product is and where it's going — no horizon collapsed into a feature list.

H1 · R&D NOW
MVP — testable Dec/Jan
  • POS integration — read-only sync with history backfill
  • Recipe graph — dish to ingredients to cost
  • Demand forecast in shadow mode
  • Misa list with reasoning shown
  • Recipe-locked inventory deduction
  • Draft purchase cart per supplier
  • TES labor cost view
H2 · YEAR 1
Post soft-launch
  • One-tap supplier ordering via integrations
  • Prep sequencing by station and time
  • Waste vs theoretical variance tracking
  • Demand-driven shift scheduling (TES-legal)
  • Payroll export to Netvisor / Procountor
  • Menu engineering — margin and velocity
  • Oiva food-safety compliance export
H3 · YEAR 2–3
Data flywheel activates
  • Network cold-start — new kitchen gets a working forecast on day one
  • Supplier price intelligence from pooled invoice data
  • Autonomous replenishment for stable ingredients
  • Multi-market compliance plug-ins — Sweden, Estonia, UK
  • Trend radar — rising dishes in your area
  • Revenue simulation — "what if we open Mondays?"
For those who want to understand why

Features are clonable. This isn't.

Every H1 feature was chosen because it captures data that makes an H3 feature possible. The loop isn't a product decision — it's a data accumulation strategy. A competitor who starts building tomorrow cannot buy three years of per-kitchen learned patterns, override histories, and supplier price data. That's the moat. It compounds with time.

"Finland first because TES is one of the hardest labor compliance regimes in the Nordics. If the architecture works here, adding Sweden or Estonia is a configuration project — not a rewrite."
The CodFleet ecosystem

Orbis is the first layer. The others sharpen it.

CODOrbis
Business intelligence
CODGo
Demand · bookings
CODAlly
Workforce · shifts
R&D cohort — limited places

Apply for early access.

We're onboarding a deliberate cohort of independent restaurants for the R&D phase. Not chains. Not franchises. Kitchens where the owner is still in the room — because that's where the real decisions happen and the real feedback comes from.

  • Independent restaurant — single location preferred
  • Owner-operated or owner-present daily
  • Existing POS system (any major provider)
  • Willing to share operational feedback during R&D
  • Helsinki preferred — other cities considered

R&D access is selective and free during testing. We'll follow up within 48 hours.