.webp)
ROLE
-
Led the end-to-end design of Venngage’s AI — from vision and strategy to design execution.
-
Defined how AI supports creation, editing, and reuse, and partnered with engineering to align UX with system capabilities.
TEAM
-
1 Product Designer (me)
-
3 Engineers
-
1 Engineering Manager
-
1 QA
From a One-Click Generator into a Trusted Partner
The AI generator promised a magical and effortless creation, but without quality, visibility, or control, users lost trust.
I reimagined how AI supports the creative process, evolving it through three phases into a continuous, collaborative system:
Phase 1 – Guided user creation through transparency and clarity.
Phase 2 – Embedded AI into workflows to drive adoption.
Phase 3 – Fostered partnership and continuity through an adaptive co-pilot.
Together, these phases transformed AI from a “magic button” into a trusted creative partner that delivers real value for both users and the business.
Problem
To understand why users were dropping off, I combined quantitative data with user insights.
Mixpanel revealed an 80% drop-off, showing most users abandoned what AI created. A scoring framework measuring structure, visual relevance, and prompt alignment confirmed that most results fell short. Using Gemini and Marvin, our internal research agent, I analyzed interview feedback and found recurring frustrations: “This isn’t the layout I asked for” and “I want AI to save time, but I still want to edit myself.”
The issue wasn’t just output quality — it was a usability gap. Users didn’t want AI to replace their workflow; they wanted it to support and enhance it.
High Drop-Off Rate
-
These problems led to a 80% drop off after the first try
-
Only 6% of users activated
Low Output Quality
-
58% of outputs lacked the correct structure (eg. timeline prompts returned unordered lists)
-
43% had irrelevant or off-topic visuals
AI Misunderstood Intent
-
AI misinterpreted intent, creating wrong layout or visual
-
AI hallucinations reduced trust
Unhappy Users
-
“It’s not doing what I asked for.”
-
“I want AI to save time, but I still want to edit myself”
who are we
designing for?
Through onboarding interviews, usage data, and prompt analysis, I identified two core personas who represented our highest-value use cases:
-
Alex, an HR manager turning internal reports into clear, professional visuals.
-
Ethan, a content marketer creating fast, on-brand assets for campaigns.
Both use cases shared a common goal—communicating complex ideas clearly and efficiently—but also the same frustrations: inconsistent results, lack of control, and off-brand outputs.
Ethan Brooks
Marketing Content Creator
GOALS
NEEDS
Create quick, visually appealing infographics
-
Speed and ease of creating visuals
-
Polished, engaging output with minimal editing
-
Relevant visuals that align with content themes
Alex Rivera
Human Resource Manager
GOALS
NEEDS
Turn long-form content into quality infographics.
-
Accuracy and fidelity to detailed content
-
Clear, structured, and professional visuals
-
Flexibility to customize and maintain brand standards
How might we...
we integrate AI into user workflows to guide, adapt, and evolve with users as a trusted creative partner?
User Objective
-
Guide creation with clarity and confidence.
-
Save time by automating repetitive design tasks.
-
Co-create with AI to refine designs efficiently.
Business Objective
-
Engagement: Users actively using AI features.
-
Adoption: AI becomes part of daily creation workflows.
-
Retention: Users return to reuse or build on past work.
-
Conversion: B2B users see value that lead to upgrades.
VISION STATEMENT
Our vision is to transform AI from a one-time generator into a continuous design partner that seamlessly supports users across discovery, creation, editing, and reuse.
Today, many business professionals (HR teams, marketers, and learning specialists) struggle to integrate AI naturally into their design workflows. They see it as a one-off feature rather than a collaborative partner that supports them.
We aim to bridge that gap by designing AI experiences that guide with transparency, fit naturally into existing workflows, and collaborate through refinement and reuse.
By building trust, habit, partnership, and continuity, AI becomes an integral part of the creative process — helping users communicate complex ideas while driving higher engagement, retention, and business value.
DESIGN
Rethinking the Generator
The first opportunity was to rethink the AI generator and improve its overall quality.
Instead of making surface-level fixes, I led a workshop with engineers to break down the full generation pipeline— from prompt parsing to generation output. Together, we identified where the AI fell short and uncovered how design could bridge the gap between automation and usability.
This revealed a key insight: thoughtful UX helps improve model performance. By designing transparent, guided interactions, we can capture clearer user intent and data, leading to more accurate and reliable AI outputs.
Phase 1 - Guiding Creation with AI
It wasn’t just about output quality, the AI generator felt like a black box: unpredictable and out of users’ control. Without visibility or clarity, most lost trust and dropped off after the first attempt.
Hypothesis
If AI could guide users from idea to starting point through transparency and clarity, they would feel more confident and engaged.
Before
One-click generation produced unpredictable results with no visibility.
After
A guided flow lets users set intent, see progress, and refine outcomes.
Phase 1 - Guiding Creation with AI
I redesigned the generation flow to make it transparent and collaborative. Instead of a one-click “generate,” users now start by selecting their desired output format, and the AI recommends templates aligned with their goals. As the design builds, real-time skeleton previews provide visibility and reinforce trust. Finally, users can explore alternatives and refine results directly, giving them a greater sense of control and agency.
It wasn’t a single redesign, but a series of small, measurable improvements, each tested and refined to see how it improved confidence, trust, and engagement.
Skeleton Previews
I started by adding skeleton previews to make AI progress visible in real time. This simple change built visibility and trust, reducing early drop-offs.
Semantic Template Matching
I then added semantic template matching to better align layouts with user intent, but without enough context, the AI still relied on guesswork, producing inconsistent results.
Category Chips
To address that, I added category chips, giving users control and clearer expectations for output types.
Guided Flow
Introduced step-by-step generation flow with template alternatives, making the process collaborative, increasing engagement by 15–20%.
The result confirmed my hypothesis:
-
Engagement: 25% to 40% (+60%)
-
Completion: 8.5% to 11% (+29%)
The redesign turned AI from a black box into a transparent, guided first step, proving that collaboration, not automation, drives engagement. With trust established, the next focus was to bring AI directly into users’ everyday workflows.
Phase 2 — Embedding AI into Workflows
Even after improving AI generation, users still saw it as a separate feature rather than part of their daily workflow. They had to leave their usual creation flow to use it, which caused friction and drop-offs. This separation made AI feel disconnected from the design process and limited its long-term value.
Hypothesis
If AI integrates naturally into existing creation flows, users will see it as a helpful extension of their workflow, increasing engagement and driving sustained adoption.
Design
Instead of introducing a separate AI feature, users now began from a familiar entry point — browsing templates. When they picked one, they could prompt or upload documents, and AI automatically transformed their content into structured visuals, generated charts and graphics, and applied their brand logos and colors for a ready-to-share draft.
This approach reduced friction and helped users adopt AI as part of their process. The personalized drafts also gave an instant sense of progress, motivating them to refine and complete rather than start over.
Impact
-
Engagement: +38%
-
Adoption: +40%
-
Time to complete: 22m → 15m
-
Conversion: +25%
Phase 3 — Continuous Collaboration with AI Co-Pilot
Users loved the AI-generated first drafts, but many still struggled with manual tweaks after creation. One recurring piece of feedback was, “Can AI help me tweak my design?”
Through research and usage analysis, I uncovered several key challenges:
-
Users lack design expertise, they struggled to make their designs look polished and professional.
-
New or occasional users found the editor overwhelming and didn’t want to invest time learning complex tools.
-
Users spent too much time maintaining visual consistency across designs.
-
Users weren’t sure how to phrase effective prompts and wanted a more guided, exploratory way to use AI.
Hypothesis
I believed that if AI could assist users directly within the editing flow, it would reduce friction, speed up completion, and improve retention by making updates and reuse effortless.
Design
After generation, users now land directly in the editor where the co-pilot sits alongside them, refining content through a conversational, in-context experience. It helps refine content, generate assets, and suggest refinements in real time while keeping users in full control.
Principles
-
Assist: Help users edit and create faster.
-
Automate: Streamline repetitive layout and formatting tasks.
-
Adapt: Personalize tone, layout, and brand style.
-
Assure: Maintain transparency, context, and user control.
Tone Guidance
I defined tone and voice guidelines to make the AI feel approachable, helpful, and confident rather than robotic or overbearing. For example, when a user requests something beyond its capabilities, co-pilot might say, “I’m still learning and can’t do that yet, but here’s how you can do it manually.” This keeps the interaction supportive and guides users forward even when the AI has limits.
Why It Works
The co-pilot bridges automation and manual editing by turning refinement into a collaboration between user and AI. It keeps users in flow while maintaining clarity, control, and consistency across the experience.
-
Natural-language editing: Enables quick, conversational adjustments that feel intuitive for all users.
-
Exploratory guidance: Encourages users to learn, explore, and edit with confidence.
-
Adaptive learning: Becomes context-aware over time, remembering tone, style, and layout preferences to personalize future creations.
IMPACT
Across these three phases, the role of AI evolved from a standalone tool into a continuous creative partner.
Phase 1 built trust, as AI guided users with clarity and visibility.
Phase 2 built habit, embedding AI naturally into everyday workflows.
Phase 3 built partnership, enabling real-time collaboration and personalization.
Now, Phase 4 focuses on continuity — where AI learns from each project and guides reuse, helping users move faster and create with growing confidence over time.
LEARNINGS
Co-Creation Over Automation
It’s important to strike the right balance between automation and control, where AI takes on the heavy lifting while empowering users to stay in charge of their work.
Design AI within familiar workflows
Trust comes not only from accurate results, but from an experience that feels intuitive and predictable. Embedding AI into familiar workflows—while providing transparency, control, and clear communication—helps users build confidence and stay in flow.
Design x Engineering Collaboration
Working closely with engineers was critical to our success. I translated user needs into design requirements, while they provided technical insights that shaped strategy, balancing feasibility with user goals.
Want the full story? Contact me to learn more.