Staff who reported working beyond regular hours
Particularly to address colleagues' needs
Generative AI Assistant Based on Company Work Data
Workmate.AI
Turning invisible alignment work into a testable AI-colleague concept
People spend about 30% of their lives working, while OECD research indicates that roughly 80% of workers experience mental stress.
- Team & duration
- Royal College of Art project team · 3 months
- Company
- Royal College of Art
Project scope & backgroundOpen details
People spend about 30% of their lives working, while OECD research indicates that roughly 80% of workers experience mental stress.
We found that excessive communication for task alignment pushes core work into overtime. Workmate.AI reduces the communication burden and provides a knowledge-search bot so employees have more equal access to company information and experience.
CompanyRoyal College of Art
Role / SkillsUser ResearchUser Experience StrategyUser Experience Design
DeliverablesService ProposalUser Research InsightsUX Hi-fi PrototypeDemo
Project dateMar 2024 - Jul 2024 · 3 months
Context
Context & Background
Faster work cycles made alignment a second job
Across four years, remote collaboration, more SaaS tools, and shorter delivery cycles multiplied the coordination work surrounding core tasks.
- 01
Four years ago
Most project information and decisions stayed within a smaller set of meetings and team tools.
- 02
The Covid shift
Remote work increased online meetings, written updates, and the number of collaboration tools used every day.
- 03
Two years ago
Teams moved faster across more parallel projects, making work status and ownership harder to predict.
- 04
Now
People spend more time finding context and aligning stakeholders, while core work is pushed beyond regular hours.
Context & Background · Survey and Interview
Alignment work consumed the time intended for core work
Product designers' time spent on communication and alignment
Communication tools used across the same work
Updates and ownership were fragmented between channels
Interview participants
5 designers · 3 managers · 2 developers
User Research · Evidence
Interviews explained why the survey signals mattered
The mixed-role sample connected overtime to three recurring forms of invisible coordination work.
Alignment displaced design work
A product designer estimated that communication and alignment could consume up to 80% of the working day.
Knowledge depended on personal networks
People repeatedly searched for helpful documents and colleagues because useful organizational knowledge was difficult to discover.
Project flow was difficult to see
Managers and contributors lacked one reliable view of requests, dependencies, and overdue work across teams.
The cost reached personal time
Frequent context switching and after-hours coordination increased stress and reduced time available for life outside work.
Supporting detail1 additional section
Context & Background
Why the workplace?
Share of a lifetime spent at work
The original page uses both one-quarter and 30% in different sections.
Workers experiencing work-related or mental stress
Reported stress figure shown for South Korea
Reported stress figure shown for the United States
Reported stress figure shown for the United Kingdom
Role & Decisions
Ownership
The decisions behind the product direction
- Mandate
- Translate mixed-method research into a coherent concept, working prototype, and delivery strategy.
- My decision
- Connected communication support, knowledge retrieval, and active task tracking in one assistant.
- Trade-off
- Kept human oversight and trust visible instead of automating sensitive coordination decisions.
- What I led
- User Research · UX Strategy · UX Design. Service Proposal, User Research Insights, UX Hi-fi Prototype, Demo.
User Research · Problem Statement
Communication and alignment inefficiencies add work, increase workload, and reduce job satisfaction
Excessive and redundant alignment
In a horizontally organized company, staff can spend 80% of their time on alignment communication.
Ineffective communication with stakeholders
Updates are repeated across tools, while reports still fail to reach the people who need them.
Challenging to predict and manage tasks
Working across several projects makes it difficult to see dependencies, priorities, and overdue work.
Concern that performance will be evaluated poorly
A limited internal network creates unequal access to useful documents, experts, and organizational experience.
Research & Strategy
Hypothesis · HMW Approach
“How might we enhance communication and alignment to help designers collaborate effectively, give them more time for core tasks, improve job satisfaction, and support work-life balance?”
Hypothesis · Opportunity to Capability
Each research opportunity became a specific product capability, preserving the line from user evidence to design decision.
01 Share once
“How might we help people share an update with all relevant stakeholders without repeating the same coordination work?”
Communication Supporter
Starts conversations around urgent work, identifies relevant people, and suggests the next alignment action or meeting time.
02 Close the resource gap
“How might we reduce performance gaps caused by unequal access to documents, expertise, and organizational context?”
Knowledge Librarian
Retrieves trusted internal knowledge, finds relevant experts, and helps people access experience outside their personal network.
03 Simplify alignment
“How might we make task alignment compact, visible, and effective across several parallel projects?”
Active Tracker
Organizes projects and requests, surfaces risk and overdue work, and prioritizes tasks by importance and dependency.
Supporting detail1 additional section
User Research · Persona
Target Persona: Joey
“Frequent minor adjustments and unproductive conflicts during negotiations prevent me from focusing on design tasks, making me worry about my career growth as a designer.”
Joey, 32, is a senior product designer at a global IT company. She wants to focus on high-value design decisions and grow her career, but meetings, fragmented resources, and repeated negotiations displace core work and extend the workday.
Her desired experience is predictable work, accessible organizational knowledge, and clear stakeholder alignment without relying on a large personal network.
Product Evidence
Outcomes · Wireframe
Workmate.AI
An AI colleague that understands work history and stakeholder context, suggests relevant tasks like a teammate, and shares organizational knowledge like a manager.
Three service capabilities
Communication Supporter
Starts conversations around urgent and relevant emails, suggests alignment tasks, and recommends available meeting times through a conversational workflow.
Knowledge Librarian
Uses natural conversation to organize company knowledge, find relevant internal experts, and recommend available meeting slots for a project.
Active Tracker
Automatically organizes projects and tasks from work data, surfaces requests, alignment, and overdue work in real time, and prioritizes tasks by risk and importance.
Communication, knowledge, and tracking prototypes
Outcomes · System Map
Workmate.AI Data Flow
Authorized work data moves from connected business tools through a retrieval and governance layer before an AI model generates an answer or recommendation.
- 01
Work data sources
- ERP
- Slack
- KakaoTalk
- Jira
- Workplace
- Asana
- 02
Workmate experience
- Dashboard
- AI Knowledge Bot
- 03
Retrieval & governance
- Vector search
- Prompt engineering
- Content authority
- 04
Model processing
- Gemini
- OpenAI
- Meta
- Kakao Brain
- Naver
Outputs
- Knowledge search
- Staff search
- Risk-based task recommendation
- Alignment support
- Schedule arrangement
Demo Testing · Workmate.AI Knowledge Bot
A tangible KakaoTalk demo for end-user testing
We created a demo for the RCA service-design exhibition, where product designers gathered, using KakaoTalk as the channel and applying Communication Supporter and Knowledge Librarian UX components.
The demo stored service-design PDFs, team CVs, and schedules, used Elasticsearch BM25 to prioritize content, and called the GPT-4 API to generate answers grounded in retrieved content.
Outcomes · Demo Testing Insights
Feedback expanded the audience and clarified trust requirements
The public exhibition moved the concept beyond the original IT-company audience and made secure data connection the central product requirement.
Broader usefulness
Participants saw value for individual designers and workers in retail, finance, education, and other industries.
Individual adoption
The service could support individual designers, not only large-company teams.
Trust before automation
Users wanted an easy way to connect and protect work data before relying on AI recommendations.
Two strategic directions
- Connect data: easy connection, safe management, personalization
- Tailored engagement: subscription and enterprise cloud or on-premise delivery
Outcomes · Service Strategy
Easy connection, secure management, and meaningful personalization
Workmate.AI should be approachable for individuals while remaining secure and operable for enterprise teams.
Easy Connection
- Single Sign-On and combined information simplify external-service integration
- Immediate access to essential productivity tools
Safe & Easy Management
- Private-cloud storage with multi-region backup
- Standard prompts make knowledge bots usable by non-specialists
Personalization
- Bots adapt to each user's work data
- Context-aware suggestions and meaningful insight reports
Supporting detail1 additional section
Subscription and enterprise delivery strategy
Individuals, project teams, and companies under 100 employees
- Basic, Pro, and Premium subscription tiers
- Monthly and annual options with discounts and a free introductory tier
- Onboarding tutorials, 24/7 contact-center support, webinars, and Q&A sessions
Companies with more than 100 employees
- Enterprise cloud or on-premise delivery
- End-to-end legacy integration, continuous monitoring, and maintenance
- Account managers, employee training, technical support, and regular optimization reports
Impact
Outcomes · Our Next Step
Move from tested concept to MVP and business validation
Step 1 · Demo and prototype testing
- Compare task-search time with and without Workmate.AI
- Compare WBS creation time
- Interview additional user segments
- Turn testing evidence into product improvements
Step 2 · MVP delivery
- Finalize requirements and product policy
- Complete UX design and UXQA
- Build the smallest secure data-connection flow
- Validate reliability before broader automation
Step 3 · Business development
- Identify critical needs by industry
- Validate pricing against operating cost
- Build a sales and onboarding toolkit
- Find joint-development partners and accelerators
What We Learned
Research, public testing, and a tangible demo strengthened the proposal
Survey hypotheses became meaningful only after interviews revealed the context behind them.
The public exhibition brought feedback from people and industries outside the original scope, expanding the team's perspective.
A tangible demo produced real usability feedback that directly informed the service and business strategy.