OLIN - AI FnB business insights

OLIN - AI FnB business insights

An AI-powered experience that turns business data into actionable insights.

kiosk cover

Overview

  • Product: OLIN by ESB — ESB is software technology that provides FnB outlet operations.

  • My role: Product Designer, collaborating with UX writer, PM, Software Engineer, Quality Assurance, and Data Team.

  • Scope of this case study: Build an MVP application, AI Business Insight — sales forecasting, P&L, action recommendations (promotion, pairing menu, and fraud detection).

  • Platforms: iOS & Android, live on App Store / Google Play

OLIN turns an outlet's raw data insight into a daily read a busy owner can act and predict sales.


Background

The F&B company faced stagnant sales performance and even losses. The owner struggled to turn the raw data into actionable items; they had limited time or resources to analyze the performance data of their outlets, so they didn;t what current outlet situation and couldn’t take effective action.

Because of these problems, a business owner or the higher-level management at an FnB industry company needs business insights with potential actions on what to do to improve their business performance.


Business & User Context

Who is the target?

FnB outlet owners and high-level managers (multi-outlet supervisors) using ESB's POS ecosystem.

  • Typically time-poor, operationally hands-on, and not necessarily data-literate.

  • Confuse how to read data insights and translate them into actionable items.

Why this problem matters to the business

  • Owners already generate the data (via back-office report) but don't extract value from it — a gap between data collected and decisions made.

Constraints going in

A new merchant and a merchant that does not have transactions in the last three months will not have forecasting insight.


Problem Map

List each distinct problem surfaced (research, support tickets, sales team feedback, or founder hypothesis) with its evidence. Concept-stage is fine — mark evidence as hypothesis where real data isn't in yet.

Problem

Who it affects

Owners must manually recap sales across days/outlets to see trends

Single & multi-outlet owners

Owners don't understand raw reports (numbers without context)

Especially non-data-literate owners

No early warning on underperforming outlets/days vs. forecasted goals

Multi-outlet supervisors

P&L visibility is fragmented or delayed, so margin issues are caught late

All owners

Owners don't know what action to take even when they see a dip (e.g. which promo, which pairing)

All owners

Fraud/leakage (e.g. void abuse, discount misuse, cash discrepancies) goes undetected until it's material

Owners with staff-run outlets

Problem Statement


How Might We

The opportunity drove us to provide a reliable AI mobile app that is accessible on users' phones. Containing the main business functions to supervise the FnB outlet with easy-to-understand data insights and to suggest what the owner should check.

Long-term goals are to get this app adopted and monetized.

Decision

Problem

Design Decision

Why

No forecast vs. actual visibility

Actual-vs-forecast comparison, outlet-level, with variance flagged in plain language

Supervisors need to triage which outlet needs attention first

Fragmented P&L

Consolidated P&L view auto-pulled from transaction + cost data

Single source of truth, no manual reconciliation

Sees the dip, doesn't know the fix

Paired "Insight → Recommended Action" cards (e.g. promo suggestion, pairing menu suggestion) directly under the metric that triggered them

Closes the loop from awareness to action — this is the differentiator vs. a generic dashboard

Undetected fraud/leakage

Anomaly flags (e.g. unusual void rate, discount pattern) surfaced as alerts, not buried in reports

Owners can't audit manually at scale; system needs to surface, not just record


Site Map & Flow

Here is the main flow, with several charts that can be indicated as pages. I designed this for the first MVP phase launched in version 1.14.

We have a short timeline to develop this project within 3 months, especially since the designer only has 2 weeks for designing; starting from Dec till Feb has launched on the Appstore/Play Store.


Design System Implementation

ESB has established the design system called Pujasera, which means "Pusat Jajanan Serba Ada". We interpret it as a single sources that provide parts of all interface needs, including foundations and components. We made it from scratch in 2023, and I was part of the team that reponsible to maintain, reviewing, and publishing.

ESB has several brand themes. For Olin, I decided to pick blue as a sign of trustworthiness. Then, I imported the foundations and UI kit into the project file.


Product Outcome

Apps Icon & Splash Screen

This is a realistic character representing Olin's AI persona, designed to establish a sense of trustworthiness as an AI business partner; within the app, the character greets users with "Hello."


Onboarding

There are three unique selling point features that Olin has. Olin's character will explain with its sound in the app.


Login/Register & Verification

For both new and existing users, only input a phone number.

We have a concept of secure OTP cause we want to reduce abuse. If the user inputs the wrong OTP 3 times, they will be banned. But don’t worry, you reactive by calling our customer service.


Main Features

Forecasting vs. Goals

The core forecasting view compresses three numbers an owner actually needs to know where they current sales position, where they should end up, and what's realistically achievable into a single glance:

  • Actual: confirmed revenue so far this month

  • Goal: remaining distance to the owner's month-end target

  • Prediction: the AI-forecasted month-end outcome based on current pace

Pairing Menu

To achieve the sales prediction goals, we suggest the potential items to combine into a package or bundle, including price and quantity to sell during this period.

Promotion

Promotion is one of effective methods to aim sales goals even more increasing. OLIN recommends effective promotions to implement that are expected to increase consumer spending.

Fraud Check

Fraud variables arise from operational behaviour without realize. OLIN can detect fraud factors like removed items, void order, cash behaviour etc., so the owner can oversee potential fraud happening in the outlet and determine what to do.

Profit & Loss

One of the specific points that the F&B owner wants to know is the Profit and Loss to oversee their revenue, Cost of a Good Sales of the product, and any operational expenses flagged.

Profit and Loss section is displayed based on the period; the owner can change it based on their reference to see the P&L variables at a glance, including the net income of the company, with a percentage comparison to the previous period.


Expected Impact

In these MVP stages, we don’t have many users; we conduct a pilot with selected merchants while spreading awareness of OLIN among FnB Owners through social media, events, etc.

We expect this product to be a business partner for FnB outlet owners. Turning raw data into readable insights, recommends actions, prevents fraud, and increases their revenue by applying our recommendations, with the help of OLIN.

Create a free website with Framer, the website builder loved by startups, designers and agencies.