Created an AI-powered UX research chatbot to make expert guidance accessible 24/7.
Industry
UX Research
Project
AI Research Chatbot
Job Title
UX Research Lead
What I did:
I designed a chatbot that helps people learn UX research and design in a practical way. It uses modern large language models and is grounded in my own corpus of research notes, case studies, workshop decks, and writing. The aim is simple. Give learners a trusted mentor that explains methods, critiques artefacts, and suggests the next best step with clear references.
The Challenge
The goal was to democratise UX research knowledge by making expert guidance accessible 24/7, helping practitioners make better decisions about their research approach and methodology selection. Two groups shaped the experience: beginners who want foundations and clear examples, and practitioners who need quick refreshers, critique, and patterns they can reuse.
The key challenge was creating a chatbot that could provide grounded, trustworthy answers while maintaining the nuance and context that human experts bring to UX research guidance.
Business Goals
- •Democratise access to expert UX research knowledge
- •Create scalable learning platform for UX practitioners
- •Build community around UX research best practices
- •Establish thought leadership in UX education
User Goals
- •Learn UX research methods and best practices
- •Get expert guidance on research approach and methodology
- •Access trusted mentor available 24/7
- •Improve research quality and decision-making
Approach
My Role
I led the experience from problem framing to high fidelity design. I mapped user and business goals, designed the conversation model, curated and structured the corpus, and built prototypes to test comprehension, trust, and time to useful outcome. I've kept the design ready for engineering discovery.
Conversation Architecture
I designed a simple loop that keeps the bot predictable and useful. The core steps include triaging intent and skill level, retrieving relevant sources from the curated corpus, composing answers with real examples, offering practical actions, and logging key takeaways with reflection prompts to close the loop.
Grounded Knowledge Base
All answers are grounded in a structured corpus built from my work. Each item carries an abstract, key takeaways, and canonical references. Project artefacts are de-identified and annotated with why we chose a path and what outcomes we saw. This turns citations into something learners can actually use.
Research and Insights
Research Methodology
I conducted interviews with UX practitioners at different career stages to understand their learning challenges and preferences. I also analysed existing learning platforms and AI tools to identify gaps in the market for practical, mentor-like guidance.
Key Insight
Practitioners didn't just want information, they needed personalised guidance and practical next steps. Most were overwhelmed by generic tutorials and needed a mentor-like experience that could adapt to their specific context and skill level.
User Research Findings
Through interviews with junior and senior UX practitioners, I discovered that the biggest challenge wasn't finding information, but knowing how to apply it in real-world scenarios. Users needed contextual advice that could bridge the gap between theory and practice.
Competitive Analysis
Analysis of existing learning platforms revealed that most focus on either theoretical knowledge (losing practical relevance) or generic advice (lacking personalisation). The opportunity was to create a tool that could provide contextual, personalised guidance based on real practitioner experience.
Market Insights
The UX field's rapid evolution creates unique challenges for practitioners trying to stay current. Users need access to both foundational knowledge and cutting-edge practices, with guidance that adapts to their specific projects and career stage.
Design Process
Content Coverage
The knowledge base covers methods such as interviews, diary studies, surveys, and usability testing; synthesis patterns including affinity mapping and insight writing; design artefacts such as journey maps, IA proposals, and interaction patterns; and facilitation playbooks for workshops and stakeholder alignment.
Trust and Safety
The bot stays inside scope, avoids legal or HR advice, and names limitations. Sensitive content is masked in the transcript and redacted in stored logs. Before any action the bot summarises what it will do and asks for confirmation.
Validation and Metrics
I evaluated the prototype with short tasks and adversarial prompts. We measured time to first useful answer, grounded accuracy against the corpus, critique helpfulness, and confidence lift after a session.
Results
Multiturn Tutoring
Adaptive conversations that adjust to user skill level and goals, providing personalised learning experiences that grow with the practitioner.
Grounded Answers
All responses include inline source chips with visible citations tied to real documents, building trust through transparency.
Practical Actions
Generate test plans, critique interview scripts, and provide actionable next steps that move research work forward with one tap.
Learning Logs
Reflection prompts that turn sessions into learning logs, helping practitioners track their growth and build a personal knowledge base.