0 → 1: Designing an AI-powered loan intake system for borrowers and lenders.
INDUSTRY
Fintech
ROLE
Product Designer
DATE
2025—2026
KEY SKILLS USED
AI Interaction Design
Systems Thinking
End-to-End UX
Workflow Simplification
What is ZorroFi?
ZorroFi is a white-label platform that helps banks collect, verify, and screen loan applications before underwriting.
It ensures applications are complete, accurate, and fraud-checked before reaching lending teams.
Problem:
Design a loan application experience that:
Reduces user drop-off caused by unclear requirements
Improves application quality for lenders
Works within existing banking systems (no technical integration changes)
This meant designing for two sides at once:
Applicants who need clarity and guidance
Lenders who need complete, reliable data
Users:
Borrowers (small business loan applicants)
Small business owners applying for loans through partner banks. They often struggle with unclear document requirements and complex application steps.
Lenders (bank underwriting teams)
Bank teams responsible for reviewing applications. They need complete, accurate, and fraud-checked submissions to make faster decisions.
My role:
I worked directly with the co-founders in an early-stage startup environment to design an onboarding experience for small business loan applicants.
My responsibilities included:
Designing an AI-guided onboarding flow for step-by-step loan document submission
Simplifying the borrower experience to reduce friction and drop-off
Ensuring the UX aligned with backend requirements for validation and fraud detection
I collaborated closely with the founder to translate product requirements into an end-to-end user flow, balancing borrower usability with lender operational needs.
Problem/Challenges:
Design a loan application experience that:
Reduces user drop-off caused by unclear requirements
Improves application quality for lenders
Works within existing banking systems (no technical integration changes)
This meant designing for two sides at once:
Applicants who need clarity and guidance
Lenders who need complete, reliable data
Constraints:
This project was built under tight timelines in an early-stage startup environment, requiring rapid design decisions with limited flexibility.
System Constraints
Must integrate into existing bank systems (no workflow changes)
Required compatibility with multiple banking environments through a white-label structure
Dual-Audience Requirements
Needed to serve both borrowers (simplicity and guidance)
And lender teams (accuracy, structure, and fraud prevention)
Product Constraints
Required early fraud detection without adding friction to the user experience
Needed to maintain a lightweight onboarding flow to minimize drop-off
Key Outcomes & Results
Designed an AI-guided onboarding experience that walks applicants through structured loan submission.
Step-by-step guidance through required documents.
Real-time feedback on issues (e.g. blurry or missing files).
Clear progress tracking throughout the application process.
A more intuitive experience for applicants, and cleaner, more complete applications for lenders before underwriting even begins.
ZorroFi accepted into UC Berkeley SkyDeck Accelerator (0.4% acceptance rate)
Signed contracts with multiple banks based on the product prototype
Evolved from early concept into a platform used in real lending environments
Final Design - Prototype
01
Problem
Understanding ZorroFi's positioning within the loan process
ZorroFi is a white-label loan intake and validation platform used by banks to collect, verify, and fraud-check loan applications before loan underwriting.
I joined the project at an early stage, where the product direction was already defined. My role was to translate a complex, multi-stakeholder system into a clear and usable onboarding experience within existing banking constraints.
Exploring the Loan Intake Journey
To understand the loan application space, I mapped the end-to-end loan intake journey through discussions with the founder and secondary research into fintech onboarding and lending workflows.
This revealed a fragmented system where applications move through multiple handoffs between borrowers, banking systems, loan officers, and underwriters—with no consistent structure connecting the experience.
1
Start Loan Application
💭 "What do I need before I start?"
💬 How long will this take?
Design Challenge -> uncertainty + commitment anxiety
2
Provide Business Information
💭 "Am I answering this correctly?"
Design Challenge -> jargon, trust, comprehension
3
Upload Documents
💬 Can I trust this platform with my sensitive data?
💭 "Where do I find that?"
💭 "What if I don't have that document?"
❗ Missing files
❗ Document rejected for unknown reason
❗ *Frustration*
Design Challenge -> confusion, overwhelm, missing documents, delivering corrective feedback without frustration
4
Wait for Bank to Review
💬 Am I done?
💭 "What's happening now?"
❗ No visibility into status
Design Challenge -> visibility and reassurance
5
Provide Additional Information Needed
❗ Repeated requests
❗ Lost Momentum
Design Challenge -> maintaining momentum and context
Core Tension
Two competing needs shaped the process:
Borrowers need simplicity and guidance.
Lenders need structure and completion.
This created friction on both sides of the experience; borrower-friendly flows need to be fast and simple, while lender requirements demand detailed, structured, and exhaustive documentation.
Borrower definition of “complete”:
Loan Application Accepted
Lender definition of “complete”:
📄
Enough verified information to confidently approve or deny the loan
Key Design Constraints & Challenges Uncovered
🧍 Borrower experience constraints
Low understanding of requirements
High risk of drop-off if process is too long or complex
Confusion during long workflows
🏦 System constraints
Must work within existing bank systems
Cannot change backend infrastructure
Limited visibility into user state
📄 Data constraints
Inconsistent documents
Missing or incomplete submissions
Need for structured outputs
🔍 Trust constraints
Sensitive financial data
Fear of loan rejection or uncertainty
From my research into the problem, I uncovered a key re-frame of the problem: rather than simply improving loan onboarding UX, the challenge became designing alignment between lenders on the backend with borrowers on the front end, with different goals, constraints, and definitions of success.
From this perspective, the opportunity was to:
Reduce uncertainty for borrowers during submission
Improve structure and completeness for lenders during review
Introduce earlier validation to prevent downstream breakdowns
Work within existing banking infrastructure constraints
"How might we guide applicants through a complex loan submission process in a way that reduces drop-off for borrowers while delivering clean, fraud-validated applications to lenders?"
02
Process
Questions I asked as I approched this problem space:
How do you design a guided intake experience for users who don’t understand requirements?
How do you introduce structure without increasing friction?
How do you surface validation at the right time without overwhelming users?
How do you support both borrower clarity and lender data quality in the same flow?
How do you design within constraints of existing banking systems?
The founding team had already established several core product concepts:
An AI-driven onboarding experience
Guided document collection and validation
"Zia" as the primary AI assistant
Fraud detection integrated into the application process
My role was translating these concepts into a clear, usable onboarding experience by determining:
👉 How much information to show at once vs. progressively reveal
👉 How to balance borrower simplicity with lender requirements
👉 How to build trust when asking users for sensitive financial information
👉 How to surface validation and fraud checks without making users feel accused
👉 How to show users what stage they're at in the dual-level application process
2A — Understanding the Loan Process
I began by mapping the end-to-end loan intake journey to understand how applications move from submission to underwriting across both user and operational layers.
This revealed a dual-layer system:
The borrower-facing experience, where users submit personal and business documentation through a guided flow
The backend validation system, where documents are reviewed for completeness, accuracy, and fraud signals before underwriting
From this, I defined two parallel progress structures:
The overall loan lifecycle: Submit → Review → Finalize
The granular submission flow: step-by-step document collection and validation
Insight:
The core challenge wasn’t simply onboarding users—it was designing alignment between two systems with competing priorities: speed and simplicity for borrowers, and accuracy, completeness, and fraud prevention for lenders.
This system definition became the foundation for both the onboarding experience and the progress tracking model.
2B — Key Design Decisions
2C — Designing the Guided Experience
“Zia” was introduced by the founding team as an AI onboarding assistant. I designed how it would function as a structured intake experience for loan applicants.
The concept was translated into a guided conversational flow that replaces traditional forms with step-by-step progression through:
Identity verification
SSN / ITIN collection
Business documentation
Financial documentation
The experience was designed to progressively surface requirements, reducing cognitive load and preventing users from feeling overwhelmed at the start of the application.
I also defined key interaction behaviors:
Real-time clarification prompts when users were uncertain
Immediate feedback for issues such as blurry or incomplete document uploads
In-chat guidance to help users locate required information
Key insight:
Zia functions as a structured intake system, not just a conversational interface—turning loan onboarding into guided progression rather than form completion.
2D — Progress and Clarity System
I designed a dual-layer progress system to address one of the main failure points in loan applications: loss of context during multi-step processes.
The experience included:
A high-level journey indicator showing overall loan status: Submit → Review → Finalize
A step-level progress tracker showing completion of individual document requirements
Together, these created continuous clarity across both the macro journey and micro tasks.
This ensured users always understood:
Where they are in the process
What is required next
What has already been completed
Why it matters:
Loan applications often break down when users lose visibility into progress or next steps. This system was designed to maintain momentum, reduce uncertainty, and prevent drop-off in long, multi-step submission flows.
03
Solution
A structured, AI-guided loan intake system
Designed to reduce friction for borrowers while increasing signal quality for lenders.
I designed a structured, AI-guided onboarding experience that streamlines loan intake for applicants while ensuring lenders receive complete, verified, and fraud-checked applications.
The solution transforms a traditionally complex, form-heavy process into a guided, step-by-step experience that supports users in real time while enforcing data quality and validation requirements.
Core Components:
AI-guided onboarding interface (“Zia”)
A conversational interface that guides applicants through the loan application step by step, replacing static forms with structured interaction.
Staged document submission flow
A structured intake process broken into identity, business, and financial verification stages to reduce complexity and improve clarity.
Real-time document validation
Immediate feedback on uploaded files, including detection of blur, mismatches, and missing or incomplete information.
Contextual guidance system
In-flow prompts that help users understand requirements and successfully submit correct documentation.
Dual progress tracking system
A combined view of overall loan progression (Submit → Review → Finalize) and step-level completion status for individual requirements.
White-label design system
A flexible interface layer that allows banks to adapt branding and presentation while maintaining a consistent underlying workflow.
Outcome of the System:
Together, these components create a guided intake experience that reduces borrower drop-off, improves submission accuracy, and ensures lenders receive structured, decision-ready applications earlier in the process.
04
Outcomes
Product & Design Outcomes
The onboarding experience transformed a complex loan application process into a guided, structured workflow that:
Reduced ambiguity through real-time guidance and validation
Increased visibility across multi-step application journeys
Improved the quality and completeness of submitted applications
Established a scalable foundation for deployment across multiple financial institutions
Business Traction & Validation
The product gained meaningful early traction and external validation:
Accepted into UC Berkeley SkyDeck, a highly competitive accelerator with a ~1% acceptance rate
Accepted into Berkeley's PAD-13 startup accelerator
Secured contracts and agreements with multiple banking partners
Progressed from early concept to adoption within real lending environments
05
Reflection
Complexity Behind the Curtain
Working on ZorroFi reinforced a challenge that exists in many products: business systems often require significant complexity behind the scenes, while users expect simplicity on the front end.
Designing this experience required balancing the needs of borrowers, lenders, compliance requirements, and fraud detection systems, all without increasing friction for applicants. It strengthened my belief that great UX often comes from hiding complexity, not removing it, and using technology and AI to do the heavy lifting whenever possible.
The Designer as Translator
Working on ZorroFi also deepened my experience designing for complex workflows and regulated environments. I learned how thoughtful onboarding can reduce confusion, build trust, and improve completion rates, while also supporting critical business goals. Most importantly, it reinforced my role as a designer: serving as both a user advocate and a translator between stakeholder needs and user needs to create solutions that work for everyone involved.
Thanks for reading!