STEALTH AI STARTUP • 2025
AI-powered brainstorming for innovation
ROLE
Product Designer
TEAM
Product Design
Product Strategy
Engineering
CONTRIBUTIONS
Customer Discovery
Process Frameworks
Prototyping
TIMELINE
May - July 2025
TL;DR
Context Aware AI-Powered Brainstorming Workflow
As the first product design intern, I led design for a GenAI brainstorming feature by collaborating with product, engineering, and customer experience, resulting in a launch-ready desktop functionality that helps R&D teams transform research insights into revenue-ready strategies.
INTRODUCTION
The enterprise platform helps R&D teams make sense of their research faster - organizing scattered insights, spotting opportunities, and turning ideas into strategies. Think of it as a workspace where AI helps teams go from "we learned something interesting" to "here's what we should build next."
PROBLEM
The platform already had a foundational brainstorming experience, but Amplitude metrics revealed low engagement. The team wanted to understand why adoption stalled and reimagine the experience to unlock its potential and scale usage across the platform.
(1)
AI-generated ideas felt irrelevant
The brainstorming feature generated generic ideas that ignored research context.
(2)
Hidden entry point
The brainstorm entry point was buried in the navigation, making the feature undiscoverable.
(3)
Drop-off reduced retention
Users who abandoned during ideation rarely returned. This drop-off reduced engagement.
How might we map team brainstorming workflows to design an AI experience that integrates with existing processes, preserves source context, and enables progression from insights to concepts?
HYPOTHESIS
To tackle complex user pain points systematically, I established targeted hypotheses that would inform design decisions and validate how our solutions could improve user outcomes and accelerate business growth.
COMPETITIVE LANDSCAPE
I reviewed AI brainstorming platforms - from LLMs to workflow tools - to see how teams generate, refine, and track ideas. I analyzed how users input context, iterate on suggestions, and save versions. This revealed what patterns work and where users abandon.
CUSTOMER DISCOVERY
I led the organization's first customer discovery to understand how teams brainstorm. Focusing on security-constrained environments, I mapped workflows and pain points to uncover where AI could help vs not.
EARLY USABILITY TESTING
Through interviews, we captured quantitative and qualitative data to validate design decisions. Quantitative metrics revealed what needed improvement and provided objective benchmarks to prioritize fixes.
EARLY SKETCHES
Research revealed users needed structured brainstorming that matched how they work in person: define the problem and constraints, generate ideas, then evaluate. The critical insight - users wanted AI for evaluation, not just generation. Scoring hundreds of ideas against multiple criteria created fatigue and bias. AI could help converge on the best concepts objectively.
CONCEPT TESTING
Brainstorming Hub - Users can start new brainstorms and revisit previous sessions in one centralized view.
AI PROTOTYPING
In the early stages, I built AI-powered prototypes to quickly test concepts and validate technical feasibility. This allowed me to demonstrate interactions and gather feedback. Hub - Users can start new brainstorms and revisit previous sessions in one centralized view.
FINAL DESIGNS
A core problem I uncovered was discoverability - users struggled to find, start, and revisit brainstorms due to unclear information architecture. I introduced a collapsible left navigation that houses all brainstorm sessions in one persistent place.
Source Library IntegrationUsers start here by selecting files or folders from their source library as contextual inputs. These sources inform AI-generated brainstorming outputs, ensuring suggestions are relevant to their sources.
IMPACT
I led the AI brainstorming redesign through 15+ user interviews, delivering CEO-approved, pixel-accurate specs that decreased drop-off rates by 35% and boosted user confidence scores by 25% in pilot testing.
Within 3 months, I moved the project from ambiguous direction to engineering-ready specs. Research revealed when teams prefer AI-generated ideas versus collaborative whiteboarding, and how to balance automation with user control.
I designed a modular canvas with input guardrails, output evaluation, and concept progression - creating reusable patterns that informed the platform's AI integration strategy.
REFLECTION
Design for speed
AI UX evolves faster than traditional product design. Ship early, iterate fast, and explore what's possible.
AI UX beyond chat
AI has unlocked possibilities beyond linear chats. The focus is to create dynamic and adaptive interfaces.
Dogfooding your own product
Use your own product. Test it. Break it. This reveals friction, edge cases, and opportunities.
STEALTH AI STARTUP • 2025
AI-powered brainstorming for innovation
ROLE
Product Designer
TEAM
Product Design
Product Strategy
Engineering
CONTRIBUTIONS
Customer Discovery
Process Frameworks
Prototyping
TIMELINE
May - July 2025
TL;DR
Context Aware AI-Powered Brainstorming Workflow
As the first product design intern, I led design for a GenAI brainstorming feature by collaborating with product, engineering, and customer experience, resulting in a launch-ready desktop functionality that helps R&D teams transform research insights into revenue-ready strategies.
INTRODUCTION
The enterprise platform helps R&D teams make sense of their research faster - organizing scattered insights, spotting opportunities, and turning ideas into strategies. Think of it as a workspace where AI helps teams go from "we learned something interesting" to "here's what we should build next."
PROBLEM
The platform already had a foundational brainstorming experience, but Amplitude metrics revealed low engagement. The team wanted to understand why adoption stalled and reimagine the experience to unlock its potential and scale usage across the platform.
(1)
AI-generated ideas felt irrelevant
The brainstorming feature generated generic ideas that ignored research context.
(2)
Hidden entry point
The brainstorm entry point was buried in the navigation, making the feature undiscoverable.
(3)
Drop-off reduced retention
Users who abandoned during ideation rarely returned. This drop-off reduced engagement.
How might we map team brainstorming workflows to design an AI experience that integrates with existing processes, preserves source context, and enables progression from insights to concepts?
HYPOTHESIS
To tackle complex user pain points systematically, I established targeted hypotheses that would inform design decisions and validate how our solutions could improve user outcomes and accelerate business growth.
COMPETITIVE LANDSCAPE
I reviewed AI brainstorming platforms - from LLMs to workflow tools - to see how teams generate, refine, and track ideas. I analyzed how users input context, iterate on suggestions, and save versions. This revealed what patterns work and where users abandon.
CUSTOMER DISCOVERY
I led the organization's first customer discovery to understand how teams brainstorm. Focusing on security-constrained environments, I mapped workflows and pain points to uncover where AI could help vs not.
EARLY USABILITY TESTING
Through interviews, we captured quantitative and qualitative data to validate design decisions. Quantitative metrics revealed what needed improvement and provided objective benchmarks to prioritize fixes.
EARLY SKETCHES
Research revealed users needed structured brainstorming that matched how they work in person: define the problem and constraints, generate ideas, then evaluate. The critical insight - users wanted AI for evaluation, not just generation. Scoring hundreds of ideas against multiple criteria created fatigue and bias. AI could help converge on the best concepts objectively.
CONCEPT TESTING
Brainstorming Hub - Users can start new brainstorms and revisit previous sessions in one centralized view.
AI PROTOTYPING
In the early stages, I built AI-powered prototypes to quickly test concepts and validate technical feasibility. This allowed me to demonstrate interactions and gather feedback. Hub - Users can start new brainstorms and revisit previous sessions in one centralized view.
FINAL DESIGNS
A core problem I uncovered was discoverability - users struggled to find, start, and revisit brainstorms due to unclear information architecture. I introduced a collapsible left navigation that houses all brainstorm sessions in one persistent place.
Source Library IntegrationUsers start here by selecting files or folders from their source library as contextual inputs. These sources inform AI-generated brainstorming outputs, ensuring suggestions are relevant to their sources.
IMPACT
I led the AI brainstorming redesign through 15+ user interviews, delivering CEO-approved, pixel-accurate specs that decreased drop-off rates by 35% and boosted user confidence scores by 25% in pilot testing.
Within 3 months, I moved the project from ambiguous direction to engineering-ready specs. Research revealed when teams prefer AI-generated ideas versus collaborative whiteboarding, and how to balance automation with user control.
I designed a modular canvas with input guardrails, output evaluation, and concept progression - creating reusable patterns that informed the platform's AI integration strategy.
REFLECTION
Design for speed
AI UX evolves faster than traditional product design. Ship early, iterate fast, and explore what's possible.
AI UX beyond chat
AI has unlocked possibilities beyond linear chats. The focus is to create dynamic and adaptive interfaces.
Dogfooding your own product
Use your own product. Test it. Break it. This reveals friction, edge cases, and opportunities.
STEALTH AI STARTUP • 2025
AI-powered brainstorming for innovation
ROLE
Product Designer
TEAM
Product Design
Product Strategy
Engineering
CONTRIBUTIONS
Customer Discovery
Process Frameworks
Prototyping
TIMELINE
May - July 2025
TL;DR
Context Aware AI-Powered Brainstorming Workflow
As the first product design intern, I led design for a GenAI brainstorming feature by collaborating with product, engineering, and customer experience, resulting in a launch-ready desktop functionality that helps R&D teams transform research insights into revenue-ready strategies.
INTRODUCTION
The enterprise platform helps R&D teams make sense of their research faster - organizing scattered insights, spotting opportunities, and turning ideas into strategies. Think of it as a workspace where AI helps teams go from "we learned something interesting" to "here's what we should build next."
PROBLEM
The platform already had a foundational brainstorming experience, but Amplitude metrics revealed low engagement. The team wanted to understand why adoption stalled and reimagine the experience to unlock its potential and scale usage across the platform.
(1)
AI-generated ideas felt irrelevant
The brainstorming feature generated generic ideas that ignored research context.
(2)
Hidden entry point
The brainstorm entry point was buried in the navigation, making the feature undiscoverable.
(3)
Drop-off reduced retention
Users who abandoned during ideation rarely returned. This drop-off reduced engagement.
How might we map team brainstorming workflows to design an AI experience that integrates with existing processes, preserves source context, and enables progression from insights to concepts?
HYPOTHESIS
To tackle complex user pain points systematically, I established targeted hypotheses that would inform design decisions and validate how our solutions could improve user outcomes and accelerate business growth.
COMPETITIVE LANDSCAPE
I reviewed AI brainstorming platforms - from LLMs to workflow tools - to see how teams generate, refine, and track ideas. I analyzed how users input context, iterate on suggestions, and save versions. This revealed what patterns work and where users abandon.
CUSTOMER DISCOVERY
I led the organization's first customer discovery to understand how teams brainstorm. Focusing on security-constrained environments, I mapped workflows and pain points to uncover where AI could help vs not.
EARLY USABILITY TESTING
Through interviews, we captured quantitative and qualitative data to validate design decisions. Quantitative metrics revealed what needed improvement and provided objective benchmarks to prioritize fixes.
EARLY SKETCHES
Research revealed users needed structured brainstorming that matched how they work in person: define the problem and constraints, generate ideas, then evaluate. The critical insight - users wanted AI for evaluation, not just generation. Scoring hundreds of ideas against multiple criteria created fatigue and bias. AI could help converge on the best concepts objectively.
CONCEPT TESTING
Brainstorming Hub - Users can start new brainstorms and revisit previous sessions in one centralized view.
AI PROTOTYPING
In the early stages, I built AI-powered prototypes to quickly test concepts and validate technical feasibility. This allowed me to demonstrate interactions and gather feedback. Hub - Users can start new brainstorms and revisit previous sessions in one centralized view.
FINAL DESIGNS
A core problem I uncovered was discoverability - users struggled to find, start, and revisit brainstorms due to unclear information architecture. I introduced a collapsible left navigation that houses all brainstorm sessions in one persistent place.
Source Library IntegrationUsers start here by selecting files or folders from their source library as contextual inputs. These sources inform AI-generated brainstorming outputs, ensuring suggestions are relevant to their sources.
IMPACT
I led the AI brainstorming redesign through 15+ user interviews, delivering CEO-approved, pixel-accurate specs that decreased drop-off rates by 35% and boosted user confidence scores by 25% in pilot testing.
Within 3 months, I moved the project from ambiguous direction to engineering-ready specs. Research revealed when teams prefer AI-generated ideas versus collaborative whiteboarding, and how to balance automation with user control.
I designed a modular canvas with input guardrails, output evaluation, and concept progression - creating reusable patterns that informed the platform's AI integration strategy.
REFLECTION
Design for speed
AI UX evolves faster than traditional product design. Ship early, iterate fast, and explore what's possible.
AI UX beyond chat
AI has unlocked possibilities beyond linear chats. The focus is to create dynamic and adaptive interfaces.
Dogfooding your own product
Use your own product. Test it. Break it. This reveals friction, edge cases, and opportunities.