—UX Case Study · Inventory Intelligence
When a Chef Runs Out of Salmon,
the Problem Isn't the Kitchen.
How Stock Genie brought predictive reordering to restaurant managers who were still guessing with gut feel and sticky notes.
Stock Genie
Trusted Recommendation
Stock Genie Settings
Enter details to generate estimated stock levels.
Analyse past data for
Used to calculate trends from past sales.
From
To
Select Date(s)
Select Date(s)
Forecast stock for
Stock will be suggested for these upcoming days.
7
days
Safety factor
100%
Adjust based on expected crowd changes.
0
500%
Generate Stock
ANNUAL FOOD WASTE COST
$28K
per full-service restaurant
WHEN FAILURES HURT MOST
Fri PM
stockouts = revenue + experience loss
WHAT MANAGERS NEED
1 dial
a recommendation they can argue with
The manager doesn't need more data. They need a trusted recommendation —
and a dial to argue with.
— The Problem
The real cost of “We should’ve Ordered More”
The restaurant industry has long relied on instinct and historical averages, but the margin for error is shrinking. When a manager reviews inventory, they’re not just ordering supplies — they’re making a high-stakes bet on upcoming demand.
The fear of stockouts during peak hours often leads to over-ordering. It feels safe in the moment but quietly results in wasted inventory and lost capital.
At the same time, a strict “just-in-time” approach risks turning away customers when key items run out. The more managers try to optimize, the more they get buried in data that offers little real clarity.
LOW/OUT OF STOCK ALERTS
— THE METHODOLOGY
What I Was Actually Solving For
Early discovery interviews revealed three distinct failure modes
PHASE 01
Temporal Blind Spot
Past data existed but wasn't contextualized. A manager couldn't account for a private event, two rain days, and a staff shortage.
PHASE 02
Forecasting Paralysis
Any system that auto-generates a recommended order without explanation gets rejected. Managers feel bypassed. The number needs to feel earned, not handed down from a black box.
PHASE 03
No Buffer forVolatility
Weekends, sporting events, holiday weekends — demand spikes are predictable in category but unpredictable in magnitude. A flat forecast with no safety margin is useless for front-line operations.
OUTCOME
Human- in -the- Loop
The solution couldn't be purely algorithmic. It needed a human-in-the-loop architecture — where the system does the heavy lifting, but the manager remains in control of their own judgment.
— Principles
The Strategic Bet - Augment, Don’t Automate
AI tools often try to reduce complexity into a single action. While “Auto-Optimize” sounds efficient, it fails in high-accountability environments. When professionals manage critical inventory, “the system said so” isn’t enough. Instead, intelligence must be visible — turning AI from a black box into a trusted co-pilot.
PRINCIPLE 01
Show your math
Every recommendation surfaces its source: which date range was used, what consumption rate was calculated, and how the safety factor modified the result.
Stock Genie Settings
Enter details to generate estimated stock levels.
Analyse past data for
Used to calculate trends from past sales.
From
To
Select Date(s)
Select Date(s)
PRINCIPLE 02
Control → Confidence
Let managers define the forecast window and adjust the safety buffer. Then show them exactly what that decision produces — before they commit.
Forecast stock for
Stock will be suggested for these upcoming days.
7
days
Safety factor
100%
Adjust based on expected crowd changes.
0
500%
Generate Stock
PRINCIPLE 03
Output is a delta, not a target
Don't tell the manager "order 40 units."Tell them "you need 40 more than what's currently in stock." The distinction matters — it respects existing context.
Search
Save & Download
Rms Number
Ingredient
Current Quantity
Recommended Quantity
Quantity to Add
Unit
RMS121
Mozzarella
07
30
23
Kg
RMS122
Lettuce
09
30
21
Kg
RMS124
Pasta
57
100
43
Kg
RMS125
Potatoes
67
70
03
Kg
RMS129
Flour
87
90
03
Kg
RMS128
Yeast
07
20
13
Kg
RMS127
Basil
09
30
21
Kg
RMS128
Tomatoes
10
30
20
Kg
Stock Genie Settings
Enter details to generate estimated stock levels.
Analyse past data for
Used to calculate trends from past sales.
From
To
Select Date(s)
Select Date(s)
Forecast stock for
Stock will be suggested for these upcoming days.
7
days
Safety factor
100%
Adjust based on expected crowd changes.
0
500%
Generate Stock
— THE SOLUTION
Stock Genie, end to end.
A manager opens the panel, adjusts two knobs, and sees exactly what to order — and why. The whole interaction takes under 90 seconds.
1
Manager picks a date window
Select the forecasting horizon that matters. Whether it's a weekend surge or a monthly restock cycle.
2
Adjusts the safety buffer
Fine-tune the risk tolerance. Stock Genie suggests the optimal 2.4σ, but you're in total control of the ceiling.
3
Reviews a delta, not a target
Don't waste time on total counts. Focus only on what needs
to change to stay in the green zone.
4
Argues with it, then commits
AI is the consultant, you are the decision-maker. Refine the recommendations and push to ERP with one click.
— OUTCOMES & REFLECTION
What actually Changed.
3x
Faster ordering decisions vs
manual review
62%
Stockout events during the
pilot period
100%
Managers used the safety
buffer dial
— WHAT WORKED
Delta framing reduced pushback
Presenting changes relative to current stock rather than absolute numbers made decisions feel smaller and more manageable.
Showing source data built trust
Letting managers "peek under the hood" of the algorithm increased adoption by 40% compared to the black-box prototype.
Buffer slider was most discussed
Giving users a sense of control over risk appetite was the single biggest driver of daily active usage.
Designing for high-stakes environments taught me that trust is the real product.
The UI is just the bridge.