Petavue: Turning Natural Language Into Business Intelligence
Petavue: Turning Natural Language Into Business Intelligence
Petavue: Turning Natural Language Into Business Intelligence
A conversational AI platform that helps deliver value to teams by querying, analyzing, and visualizing enterprise data to gain insights and draw conclusions.
A conversational AI platform that helps deliver value to teams by querying, analyzing, and visualizing enterprise data to gain insights and draw conclusions.
A conversational AI platform that helps deliver value to teams by querying, analyzing, and visualizing enterprise data to gain insights and draw conclusions.
Client
Client
Client
Petavue AI
Petavue AI
Petavue AI
Date
Date
Date
April 2023 – March 2024
April 2023 – March 2024
April 2023 – March 2024
Team
Team
Team
1 PM, 2 developers, 1 designer
1 PM, 2 developers, 1 designer
1 PM, 2 developers, 1 designer
Role and Contribution
Role and Contribution
Role and Contribution
UX Research, UX Design, Visual Design
UX Research, UX Design, Visual Design
UX Research, UX Design, Visual Design
Key Outcomes
Key Outcomes
Key Outcomes
Turning Design Decisions Into Results
Turning Design Decisions Into Results
Turning Design Decisions Into Results
Turning Design Decisions Into Results
87%
87%
87%
87%
Reduction in time making reports
Faster report generation
Reduction in time making reports
65%
65%
65%
65%
Team adoption within 3 months
Team adoption within 3 months
Team adoption within 3 months
60%
60%
60%
60%
Improvement in design efficiency
Improvement in design efficiency
Improvement in design efficiency
Problem
Problem
Problem
Data Was Hidden Behind Complex Tools and Skills
Data Was Hidden Behind Complex Tools and Skills
Data Was Hidden Behind Complex Tools and Skills
Data Was Hidden Behind Complex Tools and Skills
Extracting insights required specialized skills. This made data-driven decisions slow and inaccessible for many teams.
Extracting insights required specialized skills. This made data-driven decisions slow and inaccessible for many teams.
Extracting insights required specialized skills. This made data-driven decisions slow and inaccessible for many teams.
Long and Complex Queries
Long and Complex Queries
Querying data meant writing long, complex SQL code
Querying data meant writing long, complex SQL code
Querying data meant writing long, complex SQL code
Steep Learning Curve
Steep Learning Curve
Visualizing data involved mastering tools like Tableau
Visualizing data involved mastering tools like Tableau
Visualizing data involved mastering tools like Tableau
Time-Intensive Reporting
Time-Intensive Reporting
Even experienced analysts spent hours building reports
Even experienced analysts spent hours building reports
Even experienced analysts spent hours building reports
Even experienced analysts spent hours building reports
Reliance on Developers
Reliance on Developers
Non-technical users had to rely on developers to get answers
Non-technical users had to rely on developers to get answers
Non-technical users had to rely on developers to get answers
Core challenge
Core challenge
Core challenge
How could we make it possible for any team member to query, understand, and share data insights instantly, without relying on code or specialized tools?
How could we make it possible for any team member to query, understand, and share data insights instantly, without relying on code or specialized tools?
How could we make it possible for any team member to query, understand, and share data insights instantly, without relying on code or specialized tools?
How could we make it possible for any team member to query, understand, and share data insights instantly, without relying on code or specialized tools?
Goal
Goal
Goal
We Set Out to Make Data Insights As Simple As Chatting
We Set Out to Make Data Insights As Simple As Chatting
We Set Out to Make Data Insights As Simple As Chatting
We Set Out to Make Data Insights As Simple As Chatting
We aimed to build an AI assistant that could:
We aimed to build an AI assistant that could:
We aimed to build an AI assistant that could:
Natural Language Over SQL
Natural Language Over SQL
Natural Language Over SQL
Fetch and analyze enterprise data through natural language instructions
Fetch and analyze enterprise data through natural language instructions
Fetch and analyze enterprise data through natural language instructions
Removing Technical Barriers
Removing Technical Barriers
Removing Technical Barriers
Drastically reduce the need for SQL or technical expertise
Drastically reduce the need for SQL or technical expertise
Drastically reduce the need for SQL or technical expertise
Drastically reduce the need for SQL or technical expertise
Fast, Accessible Reporting
Fast, Accessible Reporting
Fast, Accessible Reporting
Enable any team member to create dashboards and reports quickly
Enable any team member to create dashboards and reports quickly
Enable any team member to create dashboards and reports quickly
Enable any team member to create dashboards and reports quickly
Solution
Solution
Solution
A Platform Where Anyone Could Ask, Analyze, and Share
A Platform Where Anyone Could Ask, Analyze, and Share
A Platform Where Anyone Could Ask, Analyze, and Share
A Platform Where Anyone Could Ask, Analyze, and Share
Natural Language Queries
Natural Language Queries
Natural Language Queries
Users could ask questions in plain English and get instant data insights, no SQL required.
Users could ask questions in plain English and get instant data insights, no SQL required.
Users could ask questions in plain English and get instant data insights, no SQL required.
Transparent Workflows
Transparent Workflows
Transparent Workflows
The system displayed real-time workflow execution so users could see how queries were processed.
The system displayed real-time workflow execution so users could see how queries were processed.
The system displayed real-time workflow execution so users could see how queries were processed.
Configurable Dashboards
Configurable Dashboards
Configurable Dashboards
Teams could create and share dashboards by adding widgets, tracking metrics, and asking follow-up questions.
Teams could create and share dashboards by adding widgets, tracking metrics, and asking follow-up questions.
Teams could create and share dashboards by adding widgets, tracking metrics, and asking follow-up questions.
Personas
Personas
Personas
Designed to Support Everyone From Analysts to Team Leads
Designed to Support Everyone From Analysts to Team Leads
Designed to Support Everyone From Analysts to Team Leads
Designed to Support Everyone From Analysts to Team Leads
Business Users
Business Users
Business Users
Non-technical employees needing quick answers
Non-technical employees needing quick answers
Non-technical employees needing quick answers
Data Analysts
Data Analysts
Data Analysts
Experts who cleaned and analyzed data daily
Experts who cleaned and analyzed data daily
Experts who cleaned and analyzed data daily
Team Leads
Team Leads
Team Leads
Managers tracking KPIs and team performance
Managers tracking KPIs and team performance
Managers tracking KPIs and team performance
Enterprise Teams
Enterprise Teams
Enterprise Teams
Groups needing shared, accessible insights
Groups needing shared, accessible insights
Groups needing shared, accessible insights
Groups needing shared, accessible insights
Groups needing shared, accessible insights
Methods
Methods
Methods
Structuring the Experience: Information Architecture and Wireframes
Structuring the Experience: Information Architecture and Wireframes
Structuring the Experience: Information Architecture and Wireframes
Structuring the Experience: Information Architecture and Wireframes
Before jumping into development, we focused on laying a strong foundation. These early explorations ensured the product stayed intuitive even as features expanded.
Before jumping into development, we focused on laying a strong foundation. These early explorations ensured the product stayed intuitive even as features expanded.
Before jumping into development, we focused on laying a strong foundation. These early explorations ensured the product stayed intuitive even as features expanded.
Information Architecture
Information Architecture
Information Architecture
We mapped out how users would move between uploading data, querying insights, and building dashboards.
We mapped out how users would move between uploading data, querying insights, and building dashboards.
We mapped out how users would move between uploading data, querying insights, and building dashboards.
Wireframing
Wireframing
Wireframing
Mid-fidelity wireframes helped us test different flows.
Mid-fidelity wireframes helped us test different flows.
Mid-fidelity wireframes helped us test different flows.
Minimum Viable Product
Minimum Viable Product
Minimum Viable Product
Our MVP Brought Core Features to Early Adopters
Our MVP Brought Core Features to Early Adopters
Our MVP Brought Core Features to Early Adopters
Our MVP Brought Core Features to Early Adopters
We tested with early adopters and gathered feedback on what worked, and what didn’t.
We tested with early adopters and gathered feedback on what worked, and what didn’t.
We tested with early adopters and gathered feedback on what worked, and what didn’t.
Connecting data sources
Connecting data sources
Connecting data sources
Connect documents, APIs, call recordings, and more.
Use multiple data types to inform queries.
Connect documents, APIs, call recordings, and more.
Use multiple data types to inform queries.
Connect documents, APIs, call recordings, and more.
Use multiple data types to inform queries.
Applications
Applications
Applications
View all applications in a central, categorized lists
Launch any app as a chatbot to explore data interactively.
View all applications in a central, categorized lists
Launch any app as a chatbot to explore data interactively.
View all applications in a central, categorized lists
Launch any app as a chatbot to explore data interactively.
Creating New Apps
Creating New Apps
Creating New Apps
Name and categorize custom apps.
Link relevant data sources.
Train apps with example prompts and questions.
Name and categorize custom apps.
Link relevant data sources.
Train apps with example prompts and questions.
Name and categorize custom apps.
Link relevant data sources.
Train apps with example prompts and questions.
Templates and Chats
Templates and Chats
Templates and Chats
Ask freeform questions within any app.
Run system-generated templates for quick insights.
Ask freeform questions within any app.
Run system-generated templates for quick insights.
Ask freeform questions within any app.
Run system-generated templates for quick insights.
Usability Testing
Usability Testing
Usability Testing
User Feedback Showed Where Simplicity Was Still Missing
User Feedback Showed Where Simplicity Was Still Missing
User Feedback Showed Where Simplicity Was Still Missing
User Feedback Showed Where Simplicity Was Still Missing
Data Sources
Data Sources
Data Sources
“I don’t want to add specific data sources every time, it takes too long.”
“I don’t want to add specific data sources every time, it takes too long.”
“I don’t want to add specific data sources every time, it takes too long.”
Templates
Templates
Templates
“I’d rather save my own instructions than use pre-made templates.”
“I’d rather save my own instructions than use pre-made templates.”
“I’d rather save my own instructions than use pre-made templates.”
Running Questions
Running Questions
Running Questions
“Why can’t I just ask a question directly without creating a whole app first?”
“Why can’t I just ask a question directly without creating a whole app first?”
“Why can’t I just ask a question directly without creating a whole app first?”
Dashboards
Dashboards
Dashboards
“If we’re already analyzing data, can we get a quick dashboard overview too?”
“If we’re already analyzing data, can we get a quick dashboard overview too?”
“If we’re already analyzing data, can we get a quick dashboard overview too?”
Minimum Viable Product
Minimum Viable Product
Minimum Viable Product
Version 1: A First Step Toward Conversational Data Access
Version 1: A First Step Toward Conversational Data Access
Version 1: A First Step Toward Conversational Data Access
Version 1: A First Step Toward Conversational Data Access
Our first released version brought the concept to life but also revealed important gaps. We used this feedback to refine and simplify the experience in the next version
Our first released version brought the concept to life but also revealed important gaps. We used this feedback to refine and simplify the experience in the next version
Our first released version brought the concept to life but also revealed important gaps. We used this feedback to refine and simplify the experience in the next version
Usability Testing
Usability Testing
Usability Testing
We Simplified, Removed Friction, and Added Clarity
We Simplified, Removed Friction, and Added Clarity
We Simplified, Removed Friction, and Added Clarity
We Simplified, Removed Friction, and Added Clarity
Replacing Overcomplicated Visuals With Clear Explanations
Replacing Overcomplicated Visuals With Clear Explanations
Replacing Overcomplicated Visuals With Clear Explanations
Problem: The flowchart showing how templates worked was too technical and confusing.
Problem: The flowchart showing how templates worked was too technical and confusing.
Problem: The flowchart showing how templates worked was too technical and confusing.
Solution: We replaced it with clear, text-based explanations that matched users’ mental models.
Solution: We replaced it with clear, text-based explanations that matched users’ mental models.
Solution: We replaced it with clear, text-based explanations that matched users’ mental models.
Removing Templates That Users Didn’t Need
Removing Templates That Users Didn’t Need
Removing Templates That Users Didn’t Need
Problem: Despite early interest, templates were barely used after launch.
Problem: Despite early interest, templates were barely used after launch.
Problem: Despite early interest, templates were barely used after launch.
Solution: We removed the dedicated templates section and focused on auto-generated suggestions instead.
Solution: We removed the dedicated templates section and focused on auto-generated suggestions instead.
Solution: We removed the dedicated templates section and focused on auto-generated suggestions instead.
Teaching the System to Understand Business Terminology
Teaching the System to Understand Business Terminology
Teaching the System to Understand Business Terminology
Problem: The system misunderstood company-specific terms.
Problem: The system misunderstood company-specific terms.
Problem: The system misunderstood company-specific terms.
Solution: We introduced a data dictionary feature so teams could define key metrics and terms, improving accuracy.
Solution: We introduced a data dictionary feature so teams could define key metrics and terms, improving accuracy.
Solution: We introduced a data dictionary feature so teams could define key metrics and terms, improving accuracy.
Reflections
Reflections
Reflections
What I Learned From Reimagining Data Access
What I Learned From Reimagining Data Access
What I Learned From Reimagining Data Access
What I Learned From Reimagining Data Access
Adaptability
Adaptability
Adaptability
The final product often diverges from the initial vision. Staying open to change was key.
The final product often diverges from the initial vision. Staying open to change was key.
The final product often diverges from the initial vision. Staying open to change was key.
Test Early and Often
Test Early and Often
Test Early and Often
Early testing saved time by validating assumptions and focusing our efforts where they mattered most.
Early testing saved time by validating assumptions and focusing our efforts where they mattered most.
Early testing saved time by validating assumptions and focusing our efforts where they mattered most.
Communicate With Stakeholders
Communicate With Stakeholders
Communicate With Stakeholders
Clear communication helped align priorities and make trade-offs transparent.
Clear communication helped align priorities and make trade-offs transparent.
Clear communication helped align priorities and make trade-offs transparent.
Iterate Continuously
Iterate Continuously
Iterate Continuously
Design is never truly “done.” We kept improving based on real user needs.
Design is never truly “done.” We kept improving based on real user needs.
Design is never truly “done.” We kept improving based on real user needs.
Thank you for reading!
Thank you for reading!
Thank you for reading!
Thank you for reading!
Go back to top
Go back to top
Go back to top
More Projects
More Projects
More Projects
IBM Z: Z Security Confidence
IBM Z: Z Security Confidence
IBM Z: Z Security Confidence
IBM Z: Z Security Confidence
IBM Z: Z Security Confidence
A UX research study exploring how alignment to the NIST Cybersecurity Framework drives client confidence, sales enablement, and technical clarity.
A UX research study exploring how alignment to the NIST Cybersecurity Framework drives client confidence, sales enablement, and technical clarity.
UX Research
UX Research
UX Strategy
UX Strategy
Visual Design
Visual Design
FlowScale AI: Collaborate, Deploy
FlowScale AI: Collaborate, Deploy
A collaborative platform for building and deploying ComfyUI workflows, enabling teams to create, iterate, and launch generative AI applications.
A collaborative platform for building and deploying ComfyUI workflows, enabling teams to create, iterate, and launch generative AI applications.
UX Research
UX Research
UI Design
UI Design
Visual Design
Visual Design