INTRODUCTION
User-Centric AI: Building Trust Through Transparent Design
Designing a Consent Flow & Control System Dashboard for an AI Personal Assistant
Role: Research, Strategy, UX/ UI
Duration: 3 months
Sector: AI, Privacy & Data Ethics, HCI, Consumer Technology, Responsible Innovation
PROBLEM STATEMENT
Food ain't cheap- use it before you lose it.
As AI assistants become more proactive, users start to worry:
How does it know this?
Where is this data stored?
Can I stop it from doing that again?
People want powerful automation- but only when it feels safe and transparent.
OBJECTIVES
Keep It Fresh,
Keep It Green
Designing a transparent and user-friendly consent flow:
Explain what information the AI knows and how it uses data
Let users control what it remembers
Build user trust without sacrificing the convenience of AI
APPROACH
Blending it down.
I structured the project into three key phases.
Discovery- User Research, Interviews, Competitive Analysis, Heuristic Analysis, Market Trends, Consumer Behaviors
Ideation- Strategy Development, Persona Development, Brainstorming Findings
Design- User Flows, Wireframes, Collaborative Iteration, High Fidelity Screens
DISCOVERY
No half-baked ideas.
I conducted thorough research to ensure the project was grounded in real user needs and market insights.
This began with user research, where I gathered valuable feedback directly from the target audience to identify pain points and opportunities, followed by a competitor analysis to evaluate what others in the market were doing and pinpoint areas for improvement.
I then conducted a heuristic analysis to identify usability issues in existing designs and researched market trends to gain a deeper understanding of the industry.
To better anticipate user needs, I analyzed consumer behaviors, and also engaged in stakeholder interviews to align the design strategy with business objectives. This well-rounded approach laid the foundation for a design strategy that is both user-centered and aligned with market demands.
DISCOVERY
Interviews
I spoke with 10 users across tech and non-tech backgrounds.
Key Insight: "I'm ok with AI doing things for me- but only if I know how."
COMPETITIVE ANALYSIS
Competitive Analysis
Google Assistant and Siri are vague about memory
Alexa has some memory contolls, but buried deep in settings.
ChatGPT remembers things but users are unclear about what it knows unless prompted.
USER PERSONAS
User Personas
Anna: Tech-savvy, paranoid about data and invasion of privacy
Betsy: Wants automation, but dislikes complex settings
Tom: Wants control toggles for everything
USER PERSONAS
Anna: Tech-savvy, paranoid about data and invasion of privacy
Betsy: Wants automation, but dislikes complex settings
KEY QUESTIONS
Key Questions
How can we surface memory in a way that's not overwhelming?
How do we offer control without friction?
What does "informed consent" look like in a chat-based AI?
DESIGN
Sustainable design,
effortless impact.
a) Memory Center
A central dashboard: "Here's what I remember about you based on our previous interactions", weekly memory recap sent via notifications: "Here's what I learned this week"
Grouped memories by category: Routines, Preferences, Tasks
Ont-tap options: "Forget this", "Don't track this anymore"
b) Live Consent Layer
Whenever the assistant uses remembered information, a "learned from you" badge appears.
Tapping it shows: what was remembered, when, and how it influenced the response
Option: "Don't use this again" or "Edit memory"
c) Transparency during onboarding
Let users set a trust level at the start:
Minimal memory: No personalization
Smart memory: AI remembers common actions
Full memory: AI adapts deeply to user's behavior

