Making Work Contribution Smarter with AI
Helping employees capture their work earlier, organize it better, and connect it to feedback
At Amazon, employees use Work Contribution to document work that matters for talent decisions. I designed three connected improvements: AI-generated suggestions, automated tagging, and feedback in context. Together, they increased Work Contribution creation by 62% and raised Leadership Principle tag coverage from 79% to 99.99%.
My role
UX Designer
Timeline
March–May 2026 Launches and implementation continued through July 2026
Team
1 UX Designer 1 Product Manager Engineering and Applied Science partners
Tools
Figma Kiro Claude Code
What is Work Contribution
Work Contribution is where Amazon employees keep track of important projects, actions, and impact throughout the year. Managers can use those records in Annual Review, Promotion, Contribution Summary, and Team Insights. When I joined the project, the basic manual flow was already live, but people often added their work too late, skipped important tags, and handled feedback somewhere else.
Managers need the full picture to make better decisions
Work Contribution helps employees make important work visible before talent decisions happen. Without timely, organized records, managers have to rely on memory, incomplete signals, or details collected too late in the review cycle.
A better record helps employees tell a fuller story about their impact. It also gives managers more consistent information for Annual Review, Promotion, Contribution Summary, Team Insights, and other talent programs.
My responsibility
I took over Work Contribution after the first manual experience launched, then led the next phase of the product.
01
Make documenting meaningful work easier
For employees, I wanted to make important work quicker to capture and keep contribution creation and feedback in one connected flow.
02
Increase adoption and evidence quality
For the business, I needed to grow adoption and give other talent products more complete, organized information to work with.
Employees added their work too late, and key details were often missing
Work Contribution was already being used at scale, but managers still did not have complete information when they needed it most.
01
The information came in too late
In Q1 2026, 82% of employees submitted Work Contributions after managers had already entered initial ratings. That meant managers were making early decisions without the employee input Work Contribution was meant to provide.
02
Important tags were often missing
In May 2026, only 79% of completed Work Contributions included a Leadership Principle tag.
Without those tags, other AI-powered talent products had a harder time understanding and reusing the information.
03
Feedback lived in a separate place
Employees could add collaborators in Work Contribution, but they had to switch to another product to ask for feedback. They repeated steps, and feedback providers lost the project context they needed.
Make Work Contribution easier to start, finish, and use
I treated the three features as one connected experience, with a simple goal: help employees create better Work Contributions with less effort.
Focus 1
Help employees get started
Suggest meaningful contributions from work they have already done, so they are not starting with a blank page.
Focus 2
Help employees fill in the details
Suggest Leadership Principle and Role Guideline tags automatically, while letting employees review and edit them.
Focus 3
Bring feedback into the same flow
Let employees ask collaborators for feedback while creating a Work Contribution, so everyone starts with the same project context.
The first AI idea still made people do too much work
The first concept added a “Try it out” banner that sent employees to a separate chat. It could generate content, but people still had to copy the response, go back to Work Contribution, and rebuild the draft themselves. AI was there, but the flow was still broken.
Image placeholder
Existing Work Contribution experience audit
Design approach: Keep AI in the flow
I focused on what employees were actually trying to do: create a useful Work Contribution with less effort.
Instead of sending them to another AI tool, Work Contribution could show relevant suggestions and turn the one they picked into a ready-to-edit draft.
Getting the team aligned
I worked with Product, Engineering, and Applied Science to find a realistic first launch. Using employee activity raised privacy questions, so we started with ASBX employees, where ticket activity gave us a clear, controlled source for contribution drafts. We agreed on three things:
Start with a focused pilot
Launch with ASBX employees first, then expand to a broader group.
Use ticket activity first
Build suggestions from existing ticket activity instead of adding more employee data sources.
Keep employees in control
Let AI create an editable starting point, but have employees review and publish the final contribution themselves.
Final design
The final experience brought AI suggestions directly into Work Contribution, so employees could build a draft faster.
1
Suggested contribution entry
Employees see AI-generated suggestions based on their recent work.
2
Prefilled contribution draft
When they pick a suggestion, a draft opens with the key details already filled in.
3
Review and publish
Employees edit the draft, add context, and publish when ready.
Impact
The first phase reached about 68,000 employees across seven technical job families. Early results showed that ready-to-edit suggestions helped more people capture meaningful work with less effort.
July pacing 36% above June
Creation kept growing after launch, an early sign that people were still using the experience after the initial release.
62% increase in creation
Work Contributions grew from 7,221 in May to 11,677 in June.
Manual tagging left too many contributions incomplete
Leadership Principle and Role Guideline tags help AI-powered products like Contribution Summary and Team Insights understand each Work Contribution.
But employees had to make sense of the full tag list and choose everything by hand. In May 2026, only 79% of completed Work Contributions had a Leadership Principle tag. More than one in five were missing information that other products needed.
The goal: Better tagging without taking away control
I was not trying to add as many tags as possible. I wanted to make the right tags easier to find while keeping employees in charge of how their work was described.
I designed an automated flow that recommends Leadership Principle and Role Guideline tags based on what the employee wrote.
Deciding how much AI should do
I worked with Product, Engineering, and Applied Science to decide how much control AI should have. We landed on recommendations: AI could save time, but employees would still make the final call.
Making AI suggestions clear and editable
Instead of asking employees to search the full tag list, the experience reads the contribution and suggests a smaller set of Leadership Principle and Role Guideline tags. The UI labels them as suggestions, not final decisions. Employees can remove the ones that feel wrong, add others, and confirm the final set before they publish. The flow is simpler, but the official record still belongs to the employee.
Final design
Automated Work Contribution Tagging launched on June 22, 2026. Tag suggestions now appear right in the completion flow, so employees can add the right details without searching the full list or giving up control.
1
AI-generated tag suggestions
The experience suggests relevant Leadership Principle and Role Guideline tags based on the contribution.
2
Editable tag review
Employees can remove suggestions that feel wrong, add other tags, or keep the ones they chose manually.
3
Employee confirmation
Nothing becomes final until the employee reviews and confirms it.
Impact
After launch, far more contributions included tags, and employees still used their own judgment alongside the AI suggestions.
20% increase in LP tag coverage
Leadership Principle tagging increased from 79.0% in May to 83.5% in June, reaching 99.99% in July to date.
62.5% AI tag adoption
AI-suggested tags appeared on 62.5% of July completions, up from 9.1% in May.
Feedback was separated from the work it was about
Employees could already add collaborators to a Work Contribution, but asking for feedback meant leaving the experience and opening another product.
They had to select the same people again and re-enter the project details. Feedback providers also got very little context, so the contribution and the feedback ended up as two separate records.
Design approach: Ask for feedback in the same flow
I designed a flow that lets employees ask for feedback while they add collaborators to a Work Contribution.
The project details stay attached to the request and carry through to the manager view, so no one has to enter the same information twice.
Choosing the right interaction
The main question was simple: when and where should employees ask for feedback? I explored three options:
Separate feedback button
Easy to spot, but it added clicks and competed with the main creation flow.
Request feedback after collaborator selection
The order was clear, but it added another step after employees had finished the main task.
Inline checkbox
Employees could add a collaborator and ask for feedback at the same time.
I chose the inline checkbox because it was the shortest path and still left the choice with the employee. It also connected the feedback request to the project from the start.
Add collaborator + Select feedback checkbox → Send with the contribution
Designing for 3 people in the flow
The same project context needed to work for employees, feedback providers, and managers.
For employees
Ask for project-specific feedback without leaving Work Contribution or typing the same details twice.
For feedback providers
See the contribution before responding, so it is easier to write specific, useful feedback.
For managers
See the employee’s contribution and collaborator feedback together.
Final design
The final design connected contribution creation and feedback for employees, feedback providers, and managers. Employees could ask for feedback in the same flow, and the project details stayed attached the whole way through.
1
Request feedback while adding collaborators
Employees check a box to ask specific collaborators for feedback without adding another step.
2
Respond with project context
Collaborators see the related contribution with the request, so they understand the work before they respond.
3
Review everything together
Managers see the employee’s contribution and collaborator feedback together in the same work record.
Impact
One connected experience with the right context
Bringing Work Contribution and feedback together gave everyone the same project context and removed the gaps between the two products.
Less repeated work
Employees no longer had to switch products, select collaborators again, or re-enter the project details.
More relevant feedback
Providers got clearer context about the work, which helped them write more specific responses.
A fuller picture for managers
Managers could read the employee’s own contribution next to collaborator feedback and get a more complete view of the work and its impact.
From a manual form to a smarter, connected flow
Across the three focus areas, I turned Work Contribution from a manual form into a smarter, more connected experience. Employees got help finding meaningful work, adding the right details, and gathering feedback, while staying in control of their official record.
68,000 employees reached
Phase 1 launched across seven technical job families.
62% increase in creation
Work Contribution volume increased from 7,221 in May to 11,677 in June.
99.99% LP tag coverage
With automated tagging, nearly every completed contribution included an LP tag.
Giving other talent products better information
More contributions, better tagging, and connected feedback gave products like Contribution Summary, Team Insights, Annual Review, and Promotion a fuller set of information to work with.
The project also gave us a practical pattern for using AI in sensitive talent experiences: let AI provide a starting point and save time, but keep employees responsible for reviewing and confirming the final record.
AI should help employees, not make the decision for them
The hardest part was balancing speed with trust. Work Contributions can affect important talent decisions, so making the process faster was not enough. People needed to see where AI was involved, edit anything it generated, and clearly confirm the final result themselves.
Bring the experience to more roles
Next, I would expand the suggestions beyond technical roles by adding more data sources, build in consent flows, and add Functional Competencies to the available tags.