How a data management workflow evolved into an AI analyst platform
How a data management workflow evolved into an AI analyst platform
How a data management workflow evolved into an AI analyst platform
I am an ai analyst.
I am an ai analyst.
I am an ai analyst.
Every
product
Every
product
Every
product
Every product
has A
story
has A
story
has A
story
has A story
THIS ONE IS MINE…
THIS ONE IS MINE…
THIS ONE IS MINE…
What changed when I arrived
What changed when I arrived
What changed when I arrived
Became the product's primary onboarding experience
Became the product's primary onboarding experience
Became the product's primary onboarding experience
Made ProfitOps' AI capabilities visible and tangible.
Made ProfitOps' AI capabilities visible and tangible.
Made ProfitOps' AI capabilities visible and tangible.
Became the centerpiece of product demos and walkthroughs.
Became the centerpiece of product demos and walkthroughs.
Became the centerpiece of product demos and walkthroughs.
4. Resolved a key usability gap raised during client demos.
4. Resolved a key usability gap raised during client demos.
Resolved a key usability gap raised during client demos.
What Anjali brought to the table
What Anjali brought to the table
What Anjali brought to the table
Led AI analyst experience design end-to-end
Led AI analyst experience design end-to-end
Identified flow friction through internal product observation
Identified flow friction through internal product observation
Iterated on MVP alongside PM and development team
Iterated on MVP alongside PM and development team
Designer
Anjali
Team
PM and Two developers
Timeline
8 weeks
Constraints
Limited time for extended iteration
Decisions driven by observation and product reasoning
From Obsereved Gap->
From Obsereved Gap->
From Obsereved Gap->
Client Feedback - The platform could do intelligent analysis. The interface didn't show that. it was built around how the system worked, not the value it could deliver.
Client Feedback - The platform could do intelligent analysis. The interface didn't show that. it was built around how the system worked, not the value it could deliver.
Client Feedback - The platform could do intelligent analysis. The interface didn't show that. it was built around how the system worked, not the value it could deliver.
ProductGoal
Product
Goal
ProductGoal
Build the experience around the product's most powerful capability that forecasts performance, surface signals, and explain causes.
Build the experience around the product's most powerful capability that forecasts performance, surface signals, and explain causes.
Build the experience around the product's most powerful capability that forecasts performance, surface signals, and explain causes.
Screens Before My Arrival-Old UI
Screens Before My Arrival-Old UI
Screens Before My Arrival-Old UI
The old task flow ended at Submit. No performance, no signal. To see how the data was actually performing, the user had to exit and find their way to the Business Health node on the Launchpad separately.
The old task flow ended at Submit. No performance, no signal. To see how the data was actually performing, the user had to exit and find their way to the Business Health node on the Launchpad separately.
The old task flow ended at Submit. No performance, no signal. To see how the data was actually performing, the user had to exit and find their way to the Business Health node on the Launchpad separately.
Folder based Asset Library
Folder based Asset Library
5 step wizard
This wizard was the primary experience.
Select > Preview > collect > Map > Submit
Mapping
Mapping
Flow ended
at Submit!
Viewing performance required switching screens after submission.
Submit and Save
Submit and Save
Folder based Asset Library
Upload Sample
5 step wizard
5 step wizard
This wizard was the primary experience.
Select > Preview > collect > Map > Submit
Mapping
Mapping
Flow ended
at Submit!
Flow ended
at Submit!
Viewing performance required switching screens after submission.
Submit and Save
Old UI
Data Management tool
will cover this journey in design process
Current UI
AI analyst platform
Blue steps appear in both flows — embedded inside the wizard in the current UI
Early Explorations
Early Explorations
Early Explorations
The original brief had nothing to do with AI analysts. The task was simpler - improve the existing data management experience. Early explorations stayed entirely within that frame.
The original brief had nothing to do with AI analysts. The task was simpler - improve the existing data management experience. Early explorations stayed entirely within that frame.
DIRECTION 1
Introduced domain-based selection for the first time — Product 360, Inventory 360, Pricing 360
Three-step structure — Select data 360 → Select connector → Review — was more goal-oriented
Completed the task in one flow without switching screens
Ended at "Analyse Data" — still a system action, not a business signal
Introduced domain-based selection for the first time — Product 360, Inventory 360, Pricing 360
Three-step structure — Select data 360 → Select connector → Review — was more goal-oriented
Completed the task in one flow without switching screens
Ended at "Analyse Data" — still a system action, not a business signal
Direction 2
Reduced visual complexity on the Launchpad — cleaner nodes, fewer distractions
Upload wizard remained the same — file format, preview, auto mapping, review, submit
No change in what the product was asking the user to do — just how it looked
Reduced visual complexity on the Launchpad — cleaner nodes, fewer distractions
Upload wizard remained the same — file format, preview, auto mapping, review, submit
No change in what the product was asking the user to do — just how it looked
Why Neither was enough
Why Neither was enough
Why Neither was enough
->
Business user still arrived at a system endpoint, not a business answer
Business user still arrived at a system endpoint, not a business answer
Business user still arrived at a system endpoint, not a business answer
->
Intelligent analysis, performance prediction, causal reasoning was invisible in both directions
Intelligent analysis, performance prediction, causal reasoning was invisible in both directions
Intelligent analysis, performance prediction, causal reasoning was invisible in both directions
This is where my fellow AI Analysts and I first became part of the vision.
This is where my fellow AI Analysts and I first became part of the vision.
This is where my fellow AI Analysts and I first became part of the vision.
ProfitOps already had intelligence running beneath the product. During the design exploration, inspiration from modern AI-first products revealed a new possibility: What if I worked alongside the user instead of remaining behind the product?
ProfitOps already had intelligence running beneath the product. During the design exploration, inspiration from modern AI-first products revealed a new possibility: What if I worked alongside the user instead of remaining behind the product?
ProfitOps already had intelligence running beneath the product. During the design exploration, inspiration from modern AI-first products revealed a new possibility: What if I worked alongside the user instead of remaining behind the product?
1.
Define user and their mental model
2.
What is the user actually trying to accomplish from start to finish?
3.
Define What an AI Analyst Is
4.
How should the user journey work?
5.
Updated Information Architecture
6.
Validate the interaction model screens
1.
Define user and their mental model
2.
What is the user actually trying to accomplish from start to finish?
3.
Define What an AI Analyst Is
4.
How should the user journey work?
5.
Updated Information Architecture
6.
Validate the interaction model screens
1.
Define user and their mental model
2.
What is the user actually trying to accomplish from start to finish?
3.
Define What an AI Analyst Is
4.
How should the user journey work?
5.
Updated Information Architecture
6.
Validate the interaction model screens
User Flow
Task Flow
User Flow
Task Flow
Finding the Right Flow
Finding the Right Flow
Every wireframe explored a different way to bring me into the product. The goal wasn't just to design screens, it was to build a creation journey that felt structured, intuitive, and aligned with the product vision.
Every wireframe explored a different way to bring me into the product. The goal wasn't just to design screens, it was to build a creation journey that felt structured, intuitive, and aligned with the product vision.
Direction 1
Direction 2
Direction 3
See me on Final Screens
See me on Final Screens
This is where I finally came to life. Users could now discover, create, and work with me across the product.
This is where I finally came to life. Users could now discover, create, and work with me across the product.
I kept Evolving
I kept Evolving
This wasn't the end of my journey. Every iteration refined how I worked, introduced new capabilities, and strengthened my role within the product.
This wasn't the end of my journey. Every iteration refined how I worked, introduced new capabilities, and strengthened my role within the product.
Scalable Navigation
Scalable Navigation
Introduced hierarchical navigation to organize related views and simplify exploration.
Introduced hierarchical navigation to organize related views and simplify exploration.
Introduced Objective Selection
Introduced Objective Selection
The objective step guides the AI Analyst to answer a specific business goal instead of generating generic insights.
The objective step guides the AI Analyst to answer a specific business goal instead of generating generic insights.
Preview Data Before Selection
Preview Data Before Selection
Users can expand a dataset to inspect its records without leaving the flow.
Users can expand a dataset to inspect its records without leaving the flow.
Connector Selection Experience
Connector Selection Experience
Selecting Connector displays the available enterprise integrations for the user to choose from.
Selecting Connector displays the available enterprise integrations for the user to choose from.
Watch Me Come to Life
Watch Me Come to Life
Everything we've explored so far comes together here. Watch how the final experience guides users from data to decisions through me.
Everything we've explored so far comes together here. Watch how the final experience guides users from data to decisions through me.
The Journey Continues
The Journey Continues
My story didn't end with the first release. As ProfitOps evolved, I evolved too. New requirements, customer feedback, and continuous iterations shaped how users create, configure, and collaborate with me. What you've seen here is one chapter of an experience that continues to grow.
My story didn't end with the first release. As ProfitOps evolved, I evolved too. New requirements, customer feedback, and continuous iterations shaped how users create, configure, and collaborate with me. What you've seen here is one chapter of an experience that continues to grow.
