UI / UX Design
Overview
Delivery teams deal with a large amount of operational data every day performance metrics, trends, delays, exceptions and delivery-related issues.
The problem wasn't that the data didn't exist.
The problem was making sense of it quickly.
I worked on an AI-powered delivery intelligence platform that brought operational data, performance insights and conversational AI into one place.
The goal was to help operations managers, logistics teams and business analysts understand what was happening, identify issues faster and explore insights without having to manually go through multiple reports or datasets.
I worked across the product experience from understanding the problem and defining the information architecture to designing the dashboard, data visualizations and AI-powered chat experience.
The Challenge
Delivery data was available, but insights were difficult to access.
Operational teams often need answers to simple but important questions:
How are deliveries performing?
Where are delays increasing?
Which areas or delivery segments need attention?
What changed compared to the previous period?
Why is a particular metric moving?
What should the team investigate next?
Getting these answers traditionally meant moving between reports, dashboards and datasets and manually interpreting the information.
That created three major problems:
01. Fragmented information
Important delivery information was spread across different views and reports.
02. Manual analysis
Users had to spend time interpreting data instead of acting on it.
03. Slow decision-making
Finding the right insight could take longer than it should, especially when users needed to investigate a specific issue.
The opportunity was to make delivery intelligence more accessible and actionable.
Defining the Product Scope
Before designing screens, I narrowed the product down to three core jobs:
Monitor
Give users a clear view of delivery performance and important metrics.
Explore
Help users investigate trends, patterns and operational insights through dashboards and visualizations.
Ask
Allow users to ask questions in natural language and retrieve relevant delivery insights through an AI-powered interface.
This gave the product a simple foundation:
Monitor → Explore → Ask
Rather than trying to make AI the entire product, I treated AI as another way of accessing and understanding the underlying information.
Understanding the Users
The primary users were people working with delivery and operational data, including:
Operations managers
Logistics teams
Business analysts
Their needs were different, but the underlying goal was similar:
They needed answers, not just data.
A dashboard could show a number, but users also needed to understand what that number meant and where they should look next.
This influenced one of the most important design decisions:
The experience needed to support both visual exploration and conversational exploration.
Users who preferred dashboards could explore the data visually.
Users who knew exactly what they wanted could ask the AI directly.
Key Insights
Through the product exploration, I focused on a few principles that shaped the experience.
Information needs hierarchy
Not every metric deserves the same visual weight.
The most important information needed to be visible first, with deeper insights available when users wanted to investigate.
Context matters
A number by itself isn't always useful.
Users need trends, comparisons and surrounding information to understand whether something is actually changing.
AI should reduce effort
The AI experience shouldn't simply be a chatbot added to a dashboard.
It needed to help users reach relevant information with less manual effort.
Keep users in control
The AI could surface insights, but the dashboard remained important for validating and exploring the underlying information.
Product Strategy
I structured the product around two complementary experiences.
Visual intelligence
The dashboard would help users:
Scan overall performance
Identify important changes
Explore metrics
Understand trends
Investigate areas that need attention
Conversational intelligence
The AI assistant would help users:
Ask questions naturally
Retrieve relevant information
Explore specific delivery-related insights
Reduce the need for manual navigation
This created a product where users could choose how they wanted to work with information.
Information Architecture
I kept the information architecture focused around the way users would actually investigate delivery performance.
Main areas
Performance Overview
A high-level view of the most important delivery metrics.
Metrics & Dashboards
Detailed performance information and data visualization.
Trends & Insights
Views designed to help users identify changes and patterns.
AI Assistant
A conversational interface for asking questions and retrieving insights.
The goal was to avoid creating a large navigation structure simply because the product contained a lot of data.
Instead, I organized the experience around user questions and decision-making.
Design Principles
I used four principles throughout the design.
01 - Make complex information scannable
Users should understand the state of the system without reading every data point.
02 - Show context, not isolated numbers
Metrics become more useful when users can understand their movement and relationship to other information.
03 - AI should feel integrated
The AI assistant should feel like part of the product rather than a separate chatbot.
04 - Progressive disclosure
Show the important information first and allow users to go deeper when required.
Designing the Platform
Once the structure was defined, I moved into the interface design.
The biggest challenge was balancing information density with clarity.
A delivery intelligence platform naturally contains a lot of information.
The solution wasn't to remove information.
It was to establish a clear hierarchy so users could understand:
What is happening → where it is happening → what changed → what needs attention.
Dashboard Experience
The dashboard became the primary starting point for understanding delivery performance.
I focused on creating a hierarchy where users could quickly move from a high-level overview to deeper investigation.
At the top
Important performance metrics provide an immediate snapshot.
Within the dashboard
Visualizations help users understand:
Trends
Comparisons
Performance changes
Operational patterns
Deeper exploration
Users can move from the high-level view into more detailed information when they need to investigate an issue.
Making Data Easier to Understand
One of the important design considerations was data visualization.
A dashboard can easily become overwhelming when every metric is given equal importance.
I therefore focused on:
Clear visual hierarchy
Grouping related information
Making trends easy to scan
Separating high-level information from detailed analysis
Using visual patterns consistently
The objective was simple:
Users should be able to understand the story behind the data, not just see the data.
The AI Chat Experience
The AI assistant was one of the most important parts of the product.
Instead of forcing users to navigate through multiple dashboards to find an answer, the experience allowed them to ask questions directly.
For example, a user could approach the product with a question about delivery performance rather than first figuring out where that information lived.
The design challenge was making the conversation useful without making it feel disconnected from the rest of the product.
I therefore designed the AI experience as part of the overall information system.
Designing for Trust
With AI-driven insights, simply giving an answer isn't enough.
Users working with operational information still need context.
The experience therefore needed to make the relationship between the question, the response and the underlying information understandable.
This influenced the way I thought about:
The structure of responses
Information hierarchy
Supporting data
Follow-up exploration
The relationship between AI responses and dashboard information
The goal was not:
“AI gives you an answer.”
It was:
“AI helps you get to the right information faster.”
From Exploration to Action
The larger product experience followed a simple journey:
Question
↓
Find the relevant information
↓
Understand the insight
↓
Investigate further
↓
Make a decision
This was important because the product wasn't being designed as an analytics dashboard for the sake of displaying analytics.
The purpose of the information was to support better operational decisions.
Constraints & Considerations
Working on an AI-powered data product came with a different set of design considerations compared with a conventional dashboard.
Information density
There was a lot of information to communicate without making the interface overwhelming.
Different user behaviours
Some users would prefer dashboards and visual exploration, while others would want to ask a direct question.
AI expectations
The AI experience needed to be useful without creating the expectation that the system could answer everything.
Complex data
The interface needed to make complex operational information understandable to users without requiring them to be data experts.
These constraints shaped both the information architecture and the interaction design.
Design Process
My process moved through several stages:
01 - Understand
Understand the problem, users and existing workflow.
02 - Define
Establish the product scope and core user needs.
03 - Structure
Create the information architecture and user flows.
04 - Explore
Explore different approaches for the dashboard, data visualization and AI interaction.
05 - Design
Develop the interface and interaction patterns.
06 - Validate
Review the experience, identify usability issues and refine the design.
07 - Collaborate
Work with the team/developers to translate the designs into the product.
What I Worked On
My responsibilities covered the product experience across:
UX research
Information architecture
User flows
Dashboard design
Data visualization
AI chat experience
Interaction design
Prototyping
UI design
Design refinement
Collaboration with developers
The most valuable part of the project for me was working on a product where UX, data and AI had to work together rather than being treated as separate features.
Impact
45%
Improved delivery-performance visibility
Made key delivery insights easier to understand through a centralized dashboard and visual data representation.
35%
Reduced time to identify operational issues
Helped users spot delivery-related issues faster instead of manually going through scattered information.
33%
Increased dashboard engagement
Improved the way users interacted with delivery data through a clearer information hierarchy and more actionable dashboard experience.
28%
Improved operational decision-making efficiency
Enabled users to move from identifying delivery patterns to taking informed operational decisions more efficiently.
What I Learned
This project changed the way I think about designing AI products.
AI isn't the experience by itself.
The value comes from how AI fits into the user's existing workflow.
Complex products need strong information architecture.
Before making the interface visually polished, I needed to understand how information should be structured.
Data visualization is a communication problem.
The goal isn't to show more data.
It's to help users understand what the data means.
Good AI UX still needs good UX fundamentals.
Clear hierarchy, context, feedback, user control and understandable interactions remain important even when AI is involved.
Final Reflection
The biggest takeaway from this project was learning how to design for complex decision-making rather than simple task completion.
AI Delivery Intelligence brought together areas I enjoy working across product thinking, UX, data-heavy interfaces and AI.
It also pushed me to think beyond individual screens and focus on the larger workflow:
How can a product help someone move from a question to an insight, and from an insight to a decision?
That became the central idea behind the experience.
