Back

Forkast

How can one allergy decision survive the trip from diner to kitchen?

A two-sided restaurant system that keeps evidence, questions, and safety decisions connected across four service roles.

A diner using Forkast’s personalized restaurant menu on a phone
Team
Independent project
Role
UX research + UX/UI design
Duration
Six weeks
Year
2025
  • UX Research
  • Information Architecture
  • User Flows
  • UI Design
  • Prototyping
Goal

Diners can’t trust a menu claim without asking, waiters and kitchens lose that question in the handoff, and every dish gets forced into an unsafe yes-or-no, even when the kitchen truly can’t verify it.

Method

Forkast replaces that guesswork with one shared safety record, connecting the customer-facing decision to the restaurant workflow required to verify and execute it.

Key data: findings from 11 customers

Under 2 min

Time to find a suitable dish

All 11 participants succeeded in under 2 minutes.

~50% less

Decision time vs. a traditional menu

The same group took twice as long on a traditional menu.

10 of 11: found full dish details unprompted.1 of 11 needed guidance to find the rest of the dish info.

Problem → solution overview

Three research conclusions shaped the service system and its interface states.

Unclear menu claims

Forkast dish evidence screen showing why a dish is marked safe

Show the evidence behind each safety state instead of asking diners to trust a label.

Broken staff handoffs

Forkast chef escalation queue carrying the diner’s question

Carry one question and answer record from the diner to the waiter, chef, and line cook.

Unsafe certainty

Forkast Safe, Caution, and Avoid states shown together with substitution guidance

Use Safe, Caution, and Avoid states, including an honest stop when the kitchen cannot verify a dish.

The final prototype.

One shared safety record, surfaced differently for the diner, the waiter, and the chef.

Forkast customer safety system composition showing Safe, Caution, Avoid, and substitution states
Customer safety system · final Figma composition
1 / 6
Forkast personalized menu interface
Personalized menuA scannable menu shows which dishes are safe before a diner opens the details.
Forkast safe with evidence interface
Safe with evidenceA verified state explains why the dish matches the diner’s allergy profile.
Forkast caution with context interface
Caution with contextCross-contact risk remains visible instead of being collapsed into a vague warning.
Forkast waiter dashboard interface
Waiter dashboardQuestions, tables, and ready-to-run dishes stay visible during service.
Forkast chef escalation queue interface
Chef escalation queueThe chef sees the recipe evidence and cross-contact risk behind each question.
Forkast one coherent system interface
One coherent systemThree connected interfaces present Forkast as one mobile product, not four disconnected tools.

Contributions

The work combined service research, system definition, interaction design, and interface resolution.

Learnings

Three lessons changed how the final system was framed and designed.

Research and synthesis

Framed restaurant allergy safety as a connected service problem and translated the evidence into design criteria.

01

Safety is a service-system problem.

A diner-facing filter cannot create confidence if the restaurant workflow cannot verify and execute the same decision.

System and flows

Defined the customer, waiter, chef, and line-cook relationships, including escalation and stop states.

02

Uncertainty needs a designed state.

Showing Caution, Avoid, and no-safe-option states is more responsible than forcing every dish into a confident yes or no.

UX/UI and prototyping

Designed the role-specific interfaces, reusable states, wireframes, and interactive product direction.

03

One record can support different roles.

Customers, waiters, chefs, and line cooks need different levels of detail, but they should never work from different versions of the truth.

View more projects

Continue to another UX/UI case study or return to the complete project index.