Mortgage Experiences
These are digital assets that were created for the Mortgage company project
In 2026 the Mortgage industry is massive with approximately $17.94 trillion in total outstanding mortgage balances as of mid-2025.
Discovery phase of understanding Mortgage Experiences
In 2026, I began exploring the Mortgage industry and formulated several fundamental questions.
- How can I leverage AI tools?
- What type of responses are valid?
- How can I visualize these complex responses?
Within this portfolio example, I'm going to just provide high-level details about this project. Tools, AI discussions, and Outputs to share with clients as we start the Discovery Phase.
UX Tools
- Prompt Engineering
- Vibe Coding
- AI Strategy
- JTBD
- CxDNA
- Proto-Personas
- User Flows
Moving through Discovery
To understand the Mortgage economy, it's like healthcare, and what's the problem we are trying to solve — Let's move in this a bit of a different way. So what is Discovery? It's a milestone where we are gathering open questions, creating a hypothesis to understand like the explorers of this world. In Discovery, I also find it interesting that I've practiced as a product leader over the past 10 years. Do you 'sense a thread' for product design of an emerging field's core back to these roots.
Mundo's conversation with
MS Copilot
Kicking off the conversation
- I want to create a web application within the Mortgage industry
- Persona types: Loan Officer, Loan Assistant, Loan Processor, and Borrower
- Streamline processing from borrower request to funding in 10-20 days
Moving through Prompt Engineering
My learnings over the past year have come from leveraging tools such as Claude, ChatGPT, M365, and other AI tools that use different LLMs to compare and gain familiarity with their capabilities. These are my quick tips and learnings as they relate to Microsoft (MS) Copilot, but can be used for other AI assistant tools.
5 Core notes into transcripts are important to MS Copilot to summarize and learn:
- When you start a conversation, name it to a conversation: 'Mundo's Mortgage Design Prompt'
- Tell the model, don't just sit at one level. Validate the response is aligned with stakeholder's intellectual strategy.
- Use the model for text summarization: what users need to understand and do more from the personas
AI's response
Here's where Critical thinking, a structured task, and UX thinking, is powerful in the response below, let me run through: Does this make sense? Are we missing anything? For what we need — let's move forward within the first two data story follow us here.
1. Borrower Home & Pre-Qualification
- Online account creation + identity verification (KYC)
- Credit pull + pre-qualification engine
- Pre-qualification quote for mortgage options (auto, custom)
- Automated document manager for upload of loan apps
- Integration with credit bureaus for instant pull/hard pulls
- Real-time application status updates
- E-sign collection + audit trails
2. Unified Document Management Workspace
- A single shared repository for all personas (Loan Officer, Loan Assistant, Loan Processor, Borrower)
- Processed documents classification (1003, paystub, VOE, etc)
- Real-time collaboration (assign, notify, approve)
- Version control + audit trails
- Borrower-specific upload checklists
3. Automated Task & Workflow Engine
- Designed to move files from application > underwriting > conditions > closing
- Rule-based task deployment
- Assign tasks by persona & loan types (conventional, FHA, VA, jumbo)
- SLA timers + alerts for overdue tasks
- Auto-routing to Loan Processor when documents arrive
- Condition tracking (cleared vs. outstanding)
- Exceptions logs for bottlenecks
4. Real-Time Collaboration & Communication Hub
- A cross-persona message inbox (secure and persistent)
- For notes in each processing (log develop, vendor, goals)
- Automated borrower reminders + sharing docs, signed etc, disclosed
- Multi-room smart tool reminder, overdue, scheduled, alert, waiting
- LO > Processor vs Borrower chat threads
5. Loan Officer Workstation
- Pipeline of active LO/LO team up, closing ratio
- Borrower panels + dashboard
- Auto-populated lending
- Re-approval letter generator
- Rate/program migration (or conv, fixer, scenario)
- Daily/insight tools, followup, loan queue loan comp
Mundo's summary
I believe that using any tool is not the end of a trade. The tool is only as good as the person who picks it up and learns its capabilities.
AI tools are strong, they can consolidate large amounts of data, but it's up for you, the user, to have the experience. Within Human-Centered Design making sense of complex patterns. This goes back to the stage of learning from others just to believe anymore and didn't make the thing.
The NEW goals of the AI tools are to help us Crawl, Walk, Run, and FLY. As a designer you need to keep pushing yourself to learn the fundamentals so if you don't get the foundation right your skill going to stumble. What are the fundamental forms that you will fall? Here are some items to think about: Can you skip building persona types? Can you start without understanding the problem space? What type of tools will you need to help you define toolbox? How do you define success?