(Note: Due to NDA restrictions, I’ve removed the Organization name and LOGO from this project to maintain confidentiality)
(Note: Due to NDA restrictions, I’ve removed the Organization name and LOGO from this project to maintain confidentiality)
(Note: Due to NDA restrictions, I’ve removed the Organization name and LOGO from this project to maintain confidentiality)
MY ROLE
MY ROLE
MY ROLE
Senior User Experience
Designer
Senior User Experience
Designer
Senior User Experience
Designer
Research Planning and Execution
Research Planning and Execution
Research Planning and Execution
Managing the team of 6 Designers
Managing the team of 6 Designers
Managing the team of 6 Designers
Overall 12 members (Product, AI, Engineering, UX)
Overall 12 members (Product, AI, Engineering, UX)
Overall 12 members (Product, AI, Engineering, UX)
Design language system
Design language system
Design language system
Conducting design thinking workshops
Conducting design thinking workshops
Conducting design thinking workshops
Prototyping
Prototyping
Prototyping
CHALLENGEs
CHALLENGEs
CHALLENGEs
How might we scale enterprise AI agent creation for faster delivery and Prevent manual tasks
How might we scale enterprise AI agent creation for faster delivery and Prevent manual tasks
How might we scale enterprise AI agent creation for faster delivery and Prevent manual tasks
Effort
Effort
Effort
Teams spent excessive time manually configuring AI agents with GenWizard 2.0 version through an unstructured questionnaire flow.
Teams spent excessive time manually configuring AI agents with GenWizard 2.0 version through an unstructured questionnaire flow.
Teams spent excessive time manually configuring AI agents with GenWizard 2.0 version through an unstructured questionnaire flow.
Reliability
Reliability
Reliability
Inconsistent inputs caused high error rates, repeated rework cycles, and degraded system reliability.
Inconsistent inputs caused high error rates, repeated rework cycles, and degraded system reliability.
Inconsistent inputs caused high error rates, repeated rework cycles, and degraded system reliability.
Scalability
Scalability
Scalability
Long setup times and lack of standardized templates prevented reuse and scaling across teams.
Long setup times and lack of standardized templates prevented reuse and scaling across teams.
Long setup times and lack of standardized templates prevented reuse and scaling across teams.
Why This Project Matters
Why This Project Matters
Why This Project Matters
Unstructured Agent Creation Blocking Enterprise AI Scale
Unstructured Agent Creation Blocking Enterprise AI Scale
Unstructured Agent Creation Blocking Enterprise AI Scale
Enterprise teams at Our organization relied on GenWizard Agent 2.0, a question-based workflow to create AI agents. While functional, it did not scale—agents were slow to build, inconsistent in quality, and hard to reuse. The business needed a standardized, AI-assisted system that could support enterprise-scale agent creation, faster delivery, and safe reuse across teams.
Enterprise teams at Our organization relied on GenWizard Agent 2.0, a question-based workflow to create AI agents. While functional, it did not scale—agents were slow to build, inconsistent in quality, and hard to reuse. The business needed a standardized, AI-assisted system that could support enterprise-scale agent creation, faster delivery, and safe reuse across teams.
Enterprise teams at Our organization relied on GenWizard Agent 2.0, a question-based workflow to create AI agents. While functional, it did not scale—agents were slow to build, inconsistent in quality, and hard to reuse. The business needed a standardized, AI-assisted system that could support enterprise-scale agent creation, faster delivery, and safe reuse across teams.
Information gathering
Information gathering
Information gathering
-Stakeholder interviews
-Agent 2.0 flow analysis
-User feedback mapping
-Stakeholder interviews
-Agent 2.0 flow analysis
-User feedback mapping
-Stakeholder interviews
-Agent 2.0 flow analysis
-User feedback mapping
The existing Agent 2.0 workflow presented several key challenges for both users and the business:
The existing Agent 2.0 workflow presented several key challenges for both users and the business:
The existing Agent 2.0 workflow presented several key challenges for both users and the business:
3 day workshop was conducted with Stakeholder and Team members to understand the scenario and get insights
3 day workshop was conducted with Stakeholder and Team members to understand the scenario and get insights
3 day workshop was conducted with Stakeholder and Team members to understand the scenario and get insights
The workshop uncovered key insights that shaped the product direction.
The workshop uncovered key insights that shaped the product direction.
The workshop uncovered key insights that shaped the product direction.
UX CHALLENGE
UX CHALLENGE
UX CHALLENGE
How might we design a scalable AI-assisted workflow that standardizes agent creation, provides real-time quality feedback, and enables faster, enterprise-ready agents?
How might we design a scalable AI-assisted workflow that standardizes agent creation, provides real-time quality feedback, and enables faster, enterprise-ready agents?
How might we design a scalable AI-assisted workflow that standardizes agent creation, provides real-time quality feedback, and enables faster, enterprise-ready agents?
Standardize agent creation through a structured, reusable template
Standardize agent creation through a structured, reusable template
Standardize agent creation through a structured, reusable template
Support multi-format document ingestion (PDF, DOCX, structured data).
Support multi-format document ingestion (PDF, DOCX, structured data).
Support multi-format document ingestion (PDF, DOCX, structured data).
Provide real-time feedback on completeness and content quality.
Provide real-time feedback on completeness and content quality.
Provide real-time feedback on completeness and content quality.
Enable faster creation and secure publishing of enterprise-ready agents.
Enable faster creation and secure publishing of enterprise-ready agents.
Enable faster creation and secure publishing of enterprise-ready agents.
ACTION
ACTION
ACTION
Strategic Shift: From Questionnaire to Structured Workflow
Strategic Shift: From Questionnaire to Structured Workflow
Agent 2.0 → Agent 3.0
Agent 2.0 → Agent 3.0
Agent 2.0 → Agent 3.0
I transformed the fragmented, question-based flow into a
Unified, template-driven, step-by-step creation workflow
I transformed the fragmented, question-based flow into a Unified, template-driven, step-by-step creation workflow
DEFINE
DEFINE
DEFINE
Themes and
implications
Themes and
implications
Themes and
implications
Shift in mental model
Shift in mental model
Moving users from a questionnaire-based flow to a template-driven application introduced adoption and usability challenges.
Moving users from a questionnaire-based flow to a template-driven application introduced adoption and usability challenges.
1
1
AI accuracy limitations
AI accuracy limitations
Document ingestion and data extraction were not always fully reliable across file formats, requiring careful UX design to maintain user trust.
Document ingestion and data extraction were not always fully reliable across file formats, requiring careful UX design to maintain user trust.
3
3
2
Complex validation logic
Complex validation logic
All mandatory sections had to be completed before submission, demanding clear status indicators and non-blocking error handling.
All mandatory sections had to be completed before submission, demanding clear status indicators and non-blocking error handling.
2
2
3
Flexible Standardization
Flexible Standardization
The business needed a fixed structure, while users required flexibility to upload, edit, and customize content.
The business needed a fixed structure, while users required flexibility to upload, edit, and customize content.
4
4
PRIORITIZATION MATRIX
PRIORITIZATION MATRIX
PRIORITIZATION MATRIX
Identifying touchpoints
and Prioritizing features
Identifying touchpoints
and Prioritizing features
Identifying touchpoints
and Prioritizing features
DESIGN THINKING
DESIGN THINKING
DESIGN THINKING
Brainstorm solutions
collectively
Brainstorm solutions
collectively
Brainstorm solutions
collectively
We had Representation from all teams, Product manager, AI Engineer, UX designers & tech
We had Representation from all teams, Product manager, AI Engineer, UX designers & tech
We had Representation from all teams, Product manager, AI Engineer, UX designers & tech
We conducted activities such as Affinity Mapping, Iceberg analysis, and pre-mortem sessions to uncover root causes before exploring solutions.
We conducted activities such as Affinity Mapping, Iceberg analysis, and pre-mortem sessions to uncover root causes before exploring solutions.
We conducted activities such as Affinity Mapping, Iceberg analysis, and pre-mortem sessions to uncover root causes before exploring solutions.
Conceptualization
Conceptualization
Conceptualization
I started creating the low-fi to Hi-fi concepts for primary use cases. After having a go-ahead from the Product Manager, developers, and Stakeholders on the mockups, Once we had confidence in the design, we began digitalizing designs.
I started creating the low-fi to Hi-fi concepts for primary use cases. After having a go-ahead from the Product Manager, developers, and Stakeholders on the mockups, Once we had confidence in the design, we began digitalizing designs.
I started creating the low-fi to Hi-fi concepts for primary use cases. After having a go-ahead from the Product Manager, developers, and Stakeholders on the mockups, Once we had confidence in the design, we began digitalizing designs.
Change Impact Analyzer Agent
Change Impact Analyzer Agent
Change Impact Analyzer Agent
Agent
Change Impact Analyzer Agent
Change Impact Analyzer Agent
Change Impact Analyzer Agent
Agent
QUICK TESTING & FEEDBACK
QUICK TESTING & FEEDBACK
QUICK TESTING & FEEDBACK
Test fast, fail fast,
adjust fast
Test fast, fail fast,
adjust fast
Test fast, fail fast,
adjust fast
80/ 120 user stories tested
80/ 120 user stories tested
80/ 120 user stories tested
200-250 users, 10 users across 3 countries
200-250 users, 10 users across 3 countries
200-250 users, 10 users across 3 countries
30+ Prototype
30+ Prototype
30+ Prototype
THE FINAL DESIGN
THE FINAL DESIGN
THE FINAL DESIGN
Systems Thinking & Inclusivity
Systems Thinking & Inclusivity
Systems Thinking & Inclusivity
I transformed the fragmented "Agent 2.0" experience into a unified, step-by-step "Creation Workflow" that reduced context-switching, and also applied a systems-thinking approach to the Agent 3.0 Creation Workflow.
I transformed the fragmented "Agent 2.0" experience into a unified, step-by-step "Creation Workflow" that reduced context-switching, and also applied a systems-thinking approach to the Agent 3.0 Creation Workflow.
I transformed the fragmented "Agent 2.0" experience into a unified, step-by-step "Creation Workflow" that reduced context-switching, and also applied a systems-thinking approach to the Agent 3.0 Creation Workflow.
TRADE-OFFS
TRADE-OFFS
TRADE-OFFS
Decision Making
Decision Making
Decision Making
01
01
01
Flexibility vs Standardization
Flexibility vs Standardization
Flexibility vs Standardization
Business wanted rigid structure.
Users wanted creative freedom.
Defined fixed architectural sections, but allowed free-form content within each.
Business wanted rigid structure.
Users wanted creative freedom.
Defined fixed architectural sections, but allowed free-form content within each.
Business wanted rigid structure.
Users wanted creative freedom.
Defined fixed architectural sections, but allowed free-form content within each.
02
02
02
AI Automation vs User Control
AI Automation vs User Control
AI Automation vs User Control
Full automation reduced effort but increased risk.
Human verification remained mandatory before publishing.
I prioritized reliability over speed.
Full automation reduced effort but increased risk.
Human verification remained mandatory before publishing.
I prioritized reliability over speed.
Full automation reduced effort but increased risk.
Human verification remained mandatory before publishing.
I prioritized reliability over speed.
FINAL SOLUTION
FINAL SOLUTION
FINAL SOLUTION
End to End flow
End to End flow
End to End flow
Why these changes matterED
Why these changes matterED
Why these changes matterED
The Agent Builder required a careful balance between structure and flexibility.
The Agent Builder required a careful balance between structure and flexibility.
These refinements reduced cognitive friction, and clarified AI behavior
Ensured users felt guided — not constrained — while creating enterprise agents.
These refinements reduced cognitive friction, and clarified AI behavior
Ensured users felt guided — not constrained — while creating enterprise agents.
These refinements reduced cognitive friction, and clarified AI behavior
Ensured users felt guided — not constrained — while creating enterprise agents.
FINAL REFLECTION
FINAL REFLECTION
FINAL REFLECTION
Building AI Tools that Empower Users, Not Replace Them
Building AI Tools that Empower Users, Not Replace Them
Designing for Human–AI Partnership
Designing for Human–AI Partnership
Designing for Human–AI Partnership
Building the Agent Builder taught me that a successful AI platform isn’t about showcasing the smartest model—it’s about creating a seamless partnership between the user and the tool.
Building the Agent Builder taught me that a successful AI platform isn’t about showcasing the smartest model—it’s about creating a seamless partnership between the user and the tool.
Building the Agent Builder taught me that a successful AI platform isn’t about showcasing the smartest model—it’s about creating a seamless partnership between the user and the tool.
Inclusive Design
Inclusive Design
Inclusive Design
Accessibility was integrated from the outset, ensuring WCAG AA compliance for contrast and usability.
Accessibility was integrated from the outset, ensuring WCAG AA compliance for contrast and usability.
Verified all UI elements using a color contrast checker
Design meets WCAG AA standards for normal and large text
Verified color contrast ratios (3:1 for icons, 4.5:1 for text) to ensure usability for low-vision users.
Verified all UI elements using a color contrast checker
Design meets WCAG AA standards for normal and large text
Verified color contrast ratios (3:1 for icons, 4.5:1 for text) to ensure usability for low-vision users.
Verified all UI elements using a color contrast checker
Design meets WCAG AA standards for normal and large text
Verified color contrast ratios (3:1 for icons, 4.5:1 for text) to ensure usability for low-vision users.
IMPACT
IMPACT
IMPACT
(After launch of GenWizard 3.0)
Positive Result and much more to do
(After launch of GenWizard 3.0)
Positive Result and much more to do
45%
45%
45%
Productivity Gain
Productivity Gain
Productivity Gain
Reduced average agent creation time from 5 days → 2.7 days across 18 enterprise teams
Reduced average agent creation time from 5 days → 2.7 days across 18 enterprise teams
30%
30%
30%
Validation Error Reduction
Validation Error Reduction
Validation Error Reduction
Improved quality and accuracy by
reducing workflow errors
Improved quality and accuracy by
reducing workflow errors
Improved quality and accuracy by
reducing workflow errors
40%
40%
40%
Increase in Agent Reuse
Increase in Agent Reuse
Increase in Agent Reuse
Boosted scalable adoption by increasing agent reuse across teams
Boosted scalable adoption by increasing agent reuse across teams
Boosted scalable adoption by increasing agent reuse across teams
20%
20%
20%
Drop in Workflow Abandonment
Drop in Workflow Abandonment
Drop in Workflow Abandonment
Lowered cognitive load and improved usability leading to reduction in drop-offs
Lowered cognitive load and improved usability leading to reduction in drop-offs
Lowered cognitive load and improved usability leading to reduction in drop-offs
