I find the expensive problems hiding inside large companies — and design the products that fix them.
I'm Daniel Wang. I work end-to-end across research, strategy, and interface design — currently on AI-driven and multi-platform products at a Fortune 100 insurer. I'm happiest turning a fuzzy, costly business problem into something people actually want to use.
Selected work
Six projects · 2020—2026AI Claims Predictor
A self-initiated AI prototype that predicted claim outcomes from real data — built on borrowed time, then pitched to senior leadership.
The problem
One of the company's core processes was slow and costly: claims that were mis-routed or mis-estimated drive real expense and quietly erode customer trust. I was convinced this was worth attacking, but it wasn't anyone's assigned project.
Constraints
No mandate, no budget, no headcount. I pulled together a software engineer and a data scientist on borrowed time, and we had to earn access to real, messy operational data before we could build anything credible.
Key decisions
- Started from the claims handler's day, not the model. Slower to a demo, but it meant the prediction surfaced exactly where a decision actually gets made — not as a dashboard nobody opens.
- Designed for confidence, not certainty. I showed the model's uncertainty range instead of a single hero number. Less impressive in a screenshot, but it matched how adjusters actually reason and avoided dangerous false precision.
- Scoped the prototype to one claim type. A narrower story, but a believable end-to-end one we could defend with real numbers in the room.
What I'd do differently
Bring a frontline adjuster into the room in week one rather than validating later. I designed the happy path before I'd properly pressure-tested the edge cases, and paid for it in rework.
Home Energy Management
I stood up our team's first design-concept process and shipped two validated concepts in two weeks — work a parallel team took three months to match.
The problem
The team had no repeatable way to move from a fuzzy opportunity to a concept leadership could act on. Good ideas kept dying in decks. We needed both the concepts and the machine that produces them.
Constraints
Three designers, no established process, and a two-week window — yet the output had to be credible enough to stand up in front of a company-wide audience.
Key decisions
- Built the process as we ran it. A lightweight pipeline — frame, diverge, test, narrate. Some rework along the way, but the team kept using it long after the project ended.
- Invested in the narrative, not just the artifact. I treated the handoff as a story leadership could retell themselves. Time spent on communication over extra screens, but it's why the concepts actually advanced.
What I'd do differently
Document the process while building it instead of after. Reconstructing it later cost the team roughly a week we didn't have.
Sundial
I owned the end-to-end experience for a product spanning mobile and voice — and guided other designers to ship it consistently across all three platforms.
The problem
A single product experience was fragmenting across platforms, each with its own design system. Users felt the seams, and practitioners kept reinventing the same patterns three different ways.
Constraints
Three platforms, multiple design systems to respect, and a group of practitioners working at different levels of seniority and confidence.
Key decisions
- Defined the experience at the journey level first, platform second. More upfront alignment work, but it kept the product coherent instead of three apps that merely shared a logo.
- Wrote guidance other designers could apply themselves. I gave up some personal control over every screen in exchange for quality that scaled past what I could review alone.
What I'd do differently
Invest earlier in a shared component reference. We aligned on principles before we had the concrete building blocks to make them effortless to follow.
AssignIQ
I helped take an internal operations tool from zero to one in seven months — replacing a spreadsheet-and-email workflow with a scalable web application.
The problem
Agency Recruiting relied on multiple spreadsheets and email threads to reassign an agent's book of business after they left the company. The manual process was slow, difficult to track, and vulnerable to calculation and coordination errors.
Constraints
We had seven months to build a new system around a complex operational workflow. The experience had to support large financial assignments, preserve accuracy, and move quickly enough for engineers to deliver three phases ahead of schedule.
Key decisions
- Started with day-in-the-life research. Workflow walkthroughs and impression testing exposed where analysts were doing manual math, remembering new agents, and reconciling information across files.
- Designed and tested in lockstep with engineering. Weekly sessions with users and developers replaced a traditional handoff, helping us iterate quickly across more than 60 high-fidelity screens and reusable page-level patterns.
- Centralized the decisions that mattered. Assignment data, agent search, premium splits, running totals, and status tracking moved into one interface designed for faster, more accurate decisions.
Results
The pilot validated the direction. AssignIQ is projected to process more than $5 Billion in premium in 2027, save analysts over an hour per assignment, and generate more than $155,000 in annual operational savings.
Home +
I orchestrated research, engineering, and product around a connected-home experience — and kept a sprawling effort pointed at the user.
The problem
A connected-home initiative had many capable people pulling in slightly different directions. The risk wasn't a lack of effort; it was a coherent experience getting lost between teams.
Constraints
I was the connective tissue between research, product designers, lead engineers, and technical analysts — with influence but no formal authority over any of them.
Key decisions
- Made the journey the shared artifact. I ran user-centered design and service-design workshops so the map — not any one team's backlog — became the thing everyone aligned to.
- Spent my credibility on clarity. I made the rationale behind each design decision easy to repeat, so stakeholders could carry the why without me in the room.
What I'd do differently
Set decision-making rules earlier. Alignment workshops built the shared picture, but we needed clearer ownership of the calls that picture implied.
Home Advantage
I led discovery and concept design for a centralized home-protection hub — turning scattered resources into one personalized, actionable experience backed by customer research.
The problem
Customers looking to protect and strengthen their homes had to navigate scattered articles, tools, and partner offers across the site. There was no single place that met them where they were — by home, by risk, or by what mattered most right now.
Constraints
An internal, regulated environment where the hub had to fit within the existing site architecture, taxonomy, and content ecosystem — and the experience had to be credible enough to advance through a stage gate.
Key decisions
- Started with research, not screens. A quantitative survey and moderated interviews came first, and the concept was iterated from what customers actually said — not from what we assumed they needed.
- Designed for personalization and action. Location-based content, relevant topic tags, and checklists gave clear next steps instead of generic education. Customers wanted to know what to do, not just what to know.
- Made the home explorable. I replaced static browsing with an interactive, room-by-room experience that connected risks to specific areas of the home and encouraged discovery through visual cues.
Looking ahead
Five opportunities shape the next iteration: clarify the value proposition, invest further in personalization, expand guided experiences, prioritize actionable guidance, and strengthen trust through transparency.
I'm a product designer who likes the messy front of the process — the part where the problem isn't defined yet and nobody's sure there's anything there.
Most of my work happens inside a large, regulated organization, which has taught me that good design is half craft and half persuasion. A concept only matters if the people who can fund it understand why. So I spend as much energy on the story and the stakeholders as I do on the screens.
I move fast and I write things down. I've stood up new processes, led small teams beyond my title, and shipped a working AI prototype as a side project — not because anyone asked, but because the problem was worth it.