Sketches
Prototypes, tested in the open
Small, self‑contained ideas made quickly to test a technique, answer a question, or just have fun without the scaffolding of a full project. Each sketch sets its own visual rules, so the styling inside a cell belongs to the sketch, not to this site.
Prototypes
- 01Aug 2026
Trendsetter: Visual design systems through live restyling
Visual trends are often shown through isolated examples, making it hard to understand how they behave across a full interface. This prototype applies each style to the same component set so differences in hierarchy, affordance, depth, and interaction are easier to compare.
- 02Aug 2026
User-controlled adaptation in AI responses
AI responses are typically delivered in a single structure, even when that structure may not match how a user wants to understand or act on the information. This prototype explores letting users reorganize the same response into different representations while preserving its meaning.
Nothing is sent in this demoYouI want to build a travel recommendations app but I keep going in circles. What do I actually need to think about before I start, and what should I build first?
Great question to dig into, because a travel recommendations app sounds like a weekend project and quietly turns into at least seven intertwined problems the moment you take it seriously.
Start with audience, since weekend planners who want a fast itinerary behave nothing like bucket-list researchers who browse for months, and a single interface rarely serves both well.
For data, you will want a blend of licensed points-of-interest feeds, open datasets like OpenStreetMap and Wikivoyage, and your own scraped or partnered review sources, each with different freshness and licensing terms.
On discovery UX, resist the urge to lead with a map, because most users arrive with a vague mood rather than a destination, so mood boards, season filters, and budget sliders tend to outperform pins.
Personalization should start embarrassingly simple, for example three onboarding questions about pace, budget, and company, before you even consider embeddings or collaborative filtering.
For monetization, affiliate bookings are the obvious path, but they bias recommendations toward commission-friendly inventory, so consider a paid tier that explicitly removes that bias as a selling point.
Trust and safety is where these products quietly die. Fake reviews, outdated hours, and unsafe neighborhood suggestions will burn credibility faster than any missing feature.
You will also need a moderation pipeline for user-submitted tips, because the day you launch sharing is the day someone submits a landmark that does not exist.
Launch strategy matters more than features, and a single-city deep build with genuinely great coverage beats a shallow global database every time.
Offline support is worth scoping early, since travelers are exactly the users with the worst connectivity, and cached itineraries plus offline maps change the product from novelty to tool.
Seasonality will shape everything from your data refresh cadence to your marketing calendar, so model destinations as time-varying rather than static records.
Accessibility and localization are not polish for later, because a recommendations product that only works in English or without a screen reader excludes the exact users who travel most deliberately.
Finally, measure success by trips taken rather than time in app, which is an unusual metric for consumer software but the right one for a product whose job is to get people out the door.
- 03Sep 2026
Spatial feedback for externally initiated changes
When a meeting is rescheduled by someone else, users have to reconcile the notification with their existing calendar. This prototype preserves both the original and proposed spatial states until the user responds, using motion to reinforce what changed and where.
Thursday, September 3
Product Standup9:00 AM – 9:30 AMFocus Time2:00 PM – 3:00 PMWeekly Sync4:00 PM – 4:30 PM3 schedule changes
3 awaiting you- Design CritiqueMoved by Maya Ortiz10:30 AM → 2:30 PM
- Vendor CallMoved by Dan Whitfield11:30 AM → 3:00 PM
- Project ReviewMoved by Priya Raman1:00 PM → 3:30 PM
Organizers changed these meetings.
- 04Sep 2026
User-authored adaptive commerce UI
Adaptive interfaces can reduce repeated effort, but automatic reorganization can disrupt spatial predictability. This prototype lets the system identify product information worth surfacing while the shopper decides where that personalized view should live.
BrushlessImageFrontImageKitImageDetailImageIn useGrainworks Pro Series
20V MAX Brushless 1/2 in. Hammer Drill/Driver
A compact brushless hammer drill built for daily jobsite use. Two-speed gearbox, 15-position clutch, and an all-metal chuck that holds bits square under load.
$179.00$219.00Save 18%In stock · Free pickup today
Based on your comparisons
You often look at Voltage, Weight, Battery Platform, and Warranty. Keep these together? Your placement will carry over to similar drill pages.
Product details
The Grainworks GW-2045 pairs a compact brushless motor with a two-speed transmission for controlled drilling and fastening in wood, metal, and masonry. Its balanced body keeps the tool manageable overhead while the all-metal chuck holds bits securely under load.
- Brushless motor delivers longer runtime with less maintenance
- Two-speed gearbox adapts from high-torque driving to fast drilling
- 15-position clutch helps prevent stripped fasteners
- Integrated LED illuminates the work area with a 20-second delay
Specifications
Model GW-2045- Voltage
- 20V MAX
- Chuck Size
- 1/2 in. keyless
- Weight
- 3.4 lb (bare)
- Max Torque
- 650 in-lbs
- No-Load Speed
- 0-550 / 0-2000 RPM
- Battery Platform
- PowerStack 20V
- Clutch Settings
- 15 + drill mode
- Work Light
- LED, 20s delay
- Warranty
- 3-year limited
- In the Box
- Drill, belt hook, case
- 05Sep 2026
Goal progress in AI chat
AI conversations can produce useful advice without making it clear how close the user is to actually completing a goal. This prototype explores a persistent progress model that turns a long job-search conversation into visible milestones, blockers, completed work, and next steps.
Job search workspace
YouI need to get serious about applying again. I've been putting it off because my resume is kind of outdated and my portfolio still feels unfinished.AssistantThat sounds manageable if we break it into a few concrete pieces. I'd start with four workstreams: resume, portfolio, target roles, and application workflow. We can treat interview prep as a later phase once applications are moving.YouResume is probably the easiest part. My current job is on there but the bullets are pretty generic. Portfolio is the part I keep avoiding.AssistantThen I'd use the resume as the first quick win. For the portfolio, we don't need to finish everything at once. Let's identify which projects are strong enough to publish now and which ones need work. That should make the gap feel smaller.YouI have like 3 projects I could probably use. One is old though and another is under NDA so I'd have to rewrite it pretty heavily.AssistantThat gives us a clearer picture. One project sounds close to ready, one needs reframing because of the NDA, and one may need a relevance check because of age. I'd focus first on getting two publishable case studies into acceptable shape rather than trying to perfect all three.YouYeah that makes sense. I also haven't really decided what jobs I'm applying to. Probably UX/product design but I keep seeing roles that want front-end too and that's kind of where I fit.AssistantThat's an important decision point. Before you start applying broadly, we should define your target role positioning, something like UX/Product Designer with front-end implementation experience, then use that to tune both your resume and portfolio language.YouOk. If I had to do something this week, I'd say update the resume, figure out the positioning, and get one case study cleaned up. I'm not ready to start sending applications yet.AssistantThat's a good near-term plan. I'd treat this week as your preparation phase: finalize positioning, strengthen the resume, and publish one solid case study. Once those are done, you'll be in a much better place to build a target-company list and begin applying.Static transcript. Open goal progress to see how the estimate is built.
- 06Sep 2026
Knowledge sufficiency in AI chat
Learning through AI can feel productive while still leaving important gaps hidden in the transcript. This prototype explores a persistent knowledge map that estimates what the user understands, where confidence is partial, and what they still need to learn about investing basics.
Investing basics, a learning thread
YouI've been meaning to understand investing better. I have a 401k through work but outside of that I mostly just leave money in savings because I don't really know what I'm doing.
AssistantThat's a common starting point. Before getting into individual investments, I'd make sure the foundation is clear: emergency savings, risk, diversification, account types, and the basic differences between stocks, bonds, index funds, and ETFs.
YouI know stocks are basically owning part of a company. Bonds I'm fuzzier on. And I hear index fund and ETF used almost interchangeably which confuses me.
AssistantYou've already got part of the equity side. Bonds are essentially loans to governments or companies, usually with lower expected returns and lower volatility than stocks. Index funds describe an investment strategy, tracking an index, while ETFs describe a fund structure. An ETF can be an index fund, but not every index fund is an ETF.
YouOk that actually clears up a lot. So if someone says just buy the S&P 500, they usually mean some fund that tracks it, not literally buying the index?
AssistantExactly. You can't buy the index itself. You'd buy a fund designed to track it. That's also a good point to introduce diversification, because an S&P 500 fund gives you exposure to many large U.S. companies rather than relying on a handful of individual stocks.
YouI think diversification makes sense to me. Where I get lost is like Roth IRA vs brokerage account vs my 401k. I know they're accounts but I don't really understand why I'd pick one over another.
AssistantThat's probably your biggest knowledge gap right now. The investment can be similar across those accounts, but the tax treatment, contribution rules, and access to the money differ. A 401(k) and IRA are retirement-focused accounts with tax advantages, while a taxable brokerage account is more flexible but doesn't have the same tax shelter.
YouSo I probably understand the basic investment types better than the account side. I also haven't really thought about how much risk I'm actually comfortable with.
AssistantThat's a fair assessment. You seem to have a solid introductory grasp of stocks, index funds, ETFs, and diversification. The next areas I'd focus on are account types, risk tolerance, asset allocation, fees, and how those pieces fit together into a simple long-term plan.
- 07Sep 2026
Button-to-loader transformations
Loading feedback is often shown beside or in place of the action that triggered it. This prototype explores transforming the control itself into the loading state, preserving a direct visual connection between input and system response.
Spinner
Bouncing dots
Skeleton placeholder
All button transformations ready.
- 08Sep 2026
Motus Ex Machina: Modular interface assembly through mechanical motion
This prototype explores a modular motion language for interfaces that assemble, disassemble, move, and compete for space through consistent mechanical rules. Expansion follows directional affordances, while ownership and displacement rules determine how neighboring components yield, persist, relocate, and reconstruct as the interface changes.
Modular panel collapsed.
- 09Oct 2026
Progressive commerce navigation
Traditional e-commerce often treats departments, assortments, and product detail as separate pages. This prototype explores a progressively transforming hierarchy where each level expands from the one before it, preserving spatial context as shoppers move from broad browsing to product detail.
Everyday, well considered.Collection 01