Streaming Helper
Support for the moment choosing becomes exhausting.
An HCI capstone about low-energy decision-making that later became a working browser extension and companion product.
Open live website- Role
- Research, product design, prototyping, and implementation
- Research
- 30 survey responses · 7 interviews
- Testing
- 5+ iterative prototype sessions
- Now
- Working browser extension and companion product
The case in 30 seconds
The capstone began with too many choices. Research revealed a sharper problem: hesitation when energy was low. Building it later showed something more concrete: keep trusted recommendations from disappearing before people need them.
- 01 Problem
- Repeated browsing became hesitation when users were tired.
- 02 Capstone response
- A passive helper with friend recommendations and a comfort fallback.
- 03 Product today
- A live Chrome extension and companion app for sending, saving, and returning to recommendations.
Compare titles.
Switch services.
Keep scrolling.
Effort rises while confidence drops.
Trusted suggestions make the next step smaller.
THE PRODUCT INTERVENES AT THE DECISION, NOT THE DISCOVERY
One idea, two product states
The capstone established the behavioral opportunity. Implementation forced the concept to become smaller, clearer, and technically honest.
Support hesitation across TV and computer streaming.
- Passive overlay helper
- Personalized hesitation detection
- Five friend recommendations
- Comfort Pick fallback
- Companion web app
A browser-first product with a deliberate human handoff.
- Chrome extension on supported title pages
- Send a title directly to a friend
- Saved recommendations in a web dashboard
- Friends, Comfort Picks, onboarding, and help
What survived: trusted friend recommendations, a comfort fallback, and the companion product. What changed: the interaction became user-triggered, browser-first, and feasible without pretending unvalidated detection or TV integration already worked.
The problem changed when I stopped counting options.
30 SURVEY RESPONSES
7 INTERVIEWS
AFFINITY SYNTHESIS
The recurring pattern was hesitation rather than discovery: browsing for several minutes, switching services, delaying the choice, or falling back on something familiar.
Open a platform looking for an easy way to unwind.
Scan rows, open titles, and switch between services.
Energy drops while the perceived cost of choosing rises.
Keep scrolling, leave, or return to something familiar.
HUMAN OVER ALGORITHM · FEAR OF WASTING TIME · LOW COGNITIVE ENERGY · COMFORT AS A SAFETY NET
Three decisions shaped the system
Each decision connects a research signal to a constraint and the product choice that followed.
Intervene only when support is useful.
Users were not always done choosing.
A feed or heavy recommendation tool would add another decision layer.
Keep the capstone helper passive during normal browsing, surface it at hesitation, and allow on-demand access.
Separate choosing from stopping.
Friends felt trustworthy; familiar titles worked as a safety net.
Showing both paths together would create another decision.
Friend recommendations come first during hesitation. Comfort Pick remains the fallback for when choosing is over.
Limit choice on purpose.
More options increased decision effort; too few felt restrictive.
The helper could not become another catalogue to evaluate.
Show five friend recommendations at a time: enough variety to compare, narrow enough to act.
A support system, not another streaming service
The capstone wrapped around existing services. Each part had one job, and the viewing moment stayed deliberately light.
Stayed passive during browsing, signaled at hesitation, and remained available on demand.
Managed friends, received recommendations, connected services, and the Comfort List.
Let someone send the title they were watching directly to selected friends.
Friend recommendations
Five trusted options make comparison feel smaller and less risky.
Comfort Pick
A familiar title bypasses another round of evaluation.
The working product, in context
The live product connects two moments: recommending something while it is already on screen, and returning to it when a friend is ready to watch.
LIVE PRODUCTCHROME EXTENSION + WEB APP
The extension recognizes the title without interrupting playback.
A friend is selected and the action becomes explicit.
A clear receipt closes the loop without pulling the viewer away.
Two paths appear when choosing starts to feel like work.
The helper stays compact during normal browsing. When support is wanted, a viewer can return to trusted recommendations or ask for one familiar Comfort Pick.
The companion product keeps recommendations useful later.
What a friend sends should not disappear with the viewing session. The dashboard preserves those recommendations, while the Comfort List gives familiar titles a deliberate place to live.
See the extension and companion app working together.
The walkthrough follows the complete product flow; the same key interactions are described in the screenshots above.
What the prototype established, and what it didn't
- A clearer helper symbol made passive and triggered states easier to understand.
- Friend indicators, ratings, and time cues made recommendations more useful for choosing.
- Confirmation and undo made sending a title feel recoverable rather than final.
SUCCESS WOULD BE MEASURED BY · TIME-TO-PLAY · ABANDONMENT · PLATFORM SWITCHING · RECOMMENDATION TRUST · COMFORT PICK USAGE · SATISFACTION
Building it became another round of research.
Research is useful only when it changes a product decision. The capstone’s strongest move was removing interaction until support was useful. Implementation added a second lesson: technical constraints can expose a clearer product than the original concept.
- Whether sending a recommendation is useful enough to keep the extension installed
- Where setup or permissions create hesitation
- How often saved recommendations lead to action
- Whether Comfort Picks support low-energy moments
AI accelerated synthesis, ideation, copy refinement, and rapid prototyping. I remained responsible for what to keep, remove, test, and build.