← REGISTER / ENTRY 02AMAZON IT SERVICES · SPONSORED CAPSTONE

Trust lives in the recommender, not the tool.

Amazon IT wanted more employees sharing tools across teams. Our research showed people already wanted to share. What was missing was a trusted place to do it. We designed a discovery platform, and I built a Slack bot that meets people where sharing already happens.

MY ROLE: Researcher + designer. Survey design, interviews, usability testing, synthesis. Prototyped platform. Built the Slack bot solo.
TEAM: 5 designers (V-Studio), mentored by Amazon IT's UX team
TIMELINE: 6 months · Winter/Spring 2026 · UW HCDE 593
METHODS: Survey, interviews, affinity mapping, usability testing, prototyping
TOOLS: Figma, Figma Make, Qualtrics, Slack API, Anthropic API, Railway, Claude Code

FIG. 00FINAL DESIGN · PLATFORM + SLACK BOT

300+

survey responses

interviews, ICs + managers

usability sessions

2 days

to build a working Slack bot, solo

01 · OUTCOME

STAKEHOLDER RESPONSE

Amazon IT stakeholders recommended the concept move toward implementation, and an IP-transfer review with VPs and Directors was proposed.

02 · THE PROBLEM

Amazon employees constantly find new tools and workflows, but that knowledge rarely travels beyond their own team.

WHAT LEADERSHIP ASKED

What drives tool discovery across the organization, and what gets in the way? Stakeholders also brought a hypothesis: Gen-Z employees could be the ones to spread new tools.

WHAT WE SET OUT TO LEARN

How common sharing is, and who does it
What motivates the people who share
How trust in a recommendation gets built
Where sharing actually happens day to day

03 · THE REFRAME

HOW WE GOT TO THE RIGHT QUESTION

Our survey of 300+ Amazonians tested the hypothesis directly. It didn't hold, and neither did the narrower version that followed.

STEP 1 · HYPOTHESISGen-Z employees will be the most motivated to share new tools.
STEP 2 · SURVEY, 300+94% had already shared a tool. Generation didn't predict who shares.
STEP 3 · NARROWEDStakeholders asked to split by technical vs. non-technical instead.
STEP 4 · MY CALLMy secondary research on Gen-Z sharing came up empty too. I pushed the team to stop slicing demographics and ask what motivates sharing at all.
STEP 5 · INTERVIEWS7 interviews with ICs and managers across orgs, focused on motivation and trust.

94%of respondents had shared a tool or workflow at least once

≈Sharing behavior looked similar across generations

MY PART: co-designed the survey; took part in 4 of the 7 interviews.

04 · WHAT WE FOUND

CODING + AFFINITY MAPPING

Across roles, orgs, and generations, one thread kept showing up: the barrier wasn't motivation. It was the lack of anywhere to share.

KEY INSIGHTTrust lives in the recommender, not the tool.People adopt a tool because someone they know vouches for it. So the design had to make the recommender visible, not just the tool.

FINDINGPeople are already motivated to share.

FINDINGMost sharing happens with people they already know.

FINDINGSharing already happens in Slack.

05 · WHO SHARES

4 BEHAVIORAL PROFILES

We grouped what we heard and observed into four behavioral profiles. The Nudged Contributor became our design target: the biggest untapped group.

★ DESIGN TARGETThe Nudged ContributorHas valuable knowledge but needs an external prompt to share it, such as a manager's encouragement, a direct ask, or a moment where sharing feels relevant.

PROFILE 2The ConnectorShares relationally because they know what a specific person needs, or because helping others is part of their identity.

PROFILE 3The DocumenterCreates structured, reusable knowledge artifacts (docs, demos, recordings), motivated by team efficiency and preventing knowledge loss.

PROFILE 4The ExplorerDiscovers tools constantly, shares as a byproduct of building a reputation for knowing what's out there — intrinsically motivated, reputation-aware.

06 · THE PLATFORM

DESIGNED WITH THE TEAM · FIGMA + FIGMA MAKE

One place to discover tools, see who recommends them and why, and share your own. Every feature traces back to a finding: recommenders are front and center because that's where the trust is.

6A · Home

Find tools for your role

Tools tagged by category and org-approval status, plus suggestions based on your recent tool usage, some with video walkthroughs.

↳ FROM: Amazonians are already looking for tools

FIG. 016A · HOME

6B · Tools + Workflows

See who recommends it, and how they use it

Each tool shows its recommenders and their use cases. Workflows go deeper: step-by-step guides tied to a real contributor, with files and a Q&A thread.

↳ FROM: Trust lives in the recommender

FIG. 026B · TOOLS + WORKFLOWS

6C · Analytics

Show contributors their impact

A personal dashboard: views, “found useful” counts, and how many teams and orgs a shared workflow reached.

↳ FROM: Research showed Amazonians are motivated by seeing the impact of their effort

FIG. 036C · ANALYTICS

6D · Workshops

Turn a popular guide into a live session

Contributors can host live walkthroughs of their most-requested workflows, and people register right from the workflow page.

↳ FROM: People share most with people they know

FIG. 046D · WORKSHOPS

07 · WHAT TESTING CHANGED

12 SESSIONS

WHO WE TESTED

12 Amazonians across our behavioral profiles, orgs, roles, and ages, both ICs and managers.

WHAT WE TESTED

Navigation, feature discoverability, trust signals and recommender visibility, and friction in the sharing flow.

Change 1: removed the manager analytics view

Before — Participants called it surveillance-adjacent, which would undercut the trust the platform depends on.

After — Analytics belong to the contributor, not their manager.

CHANGE 2Contributors still struggled to find how to share. We made those features easier to find, and this finding led directly to the Slack bot ↓

08 · THE SLACK BOT

MY PART · BUILT SOLO

Slack is where sharing already happened, so that's where I put the nudge.

SELECT A FEATURE TO SEE IT IN ACTION →

Built in 2 days with Claude · deployed on Railway · Slack API + Anthropic API · tested end to end in a private channel

FIG. 07NUDGE · MUTED LOOP

09 · REFLECTION

WHAT I LEARNEDPush back with data, early[2–3 sentences on redirecting stakeholders from demographics to motivation, and how you made the case]

WHAT I'D DO NEXTUsability-test the bot[The bot was tested for function, not with users. What you'd measure: do nudges lead to contributions?]

WHERE IT LANDEDRecommended for implementationAn IP-transfer review with VPs and Directors was proposed. It wasn't held during a busy period, and the project's status since then is unknown.

Team photo — V-Studio with our Amazon IT sponsors. Thank you to the Amazon IT team, particularly Michael Berg and Meiyao Li for their time and mentorship, and to my teammates Amy La, Bri Tran, Trushaa Ramanan, and Zainab Tanveer.