Designing an AI-Driven Conversational Workflow Platform
Snackk.me
AI role-play for English — built for real speaking confidence
Snackk.me turns “I should practice speaking” into something you can actually do daily — through phone-first audio role-play and YouTube-powered conversation practice.
Designed for learners who want immersive practice without the pressure of video calls — and for anyone who learns best by playing situations, not memorizing rules.
Live product • Web experience + AI role-play demos
Project Snapshot
Snackk.me is a concept-to-build product effort combining product strategy, UX, and front-end thinking — focused on a single core problem: speaking practice is the hardest habit to maintain.
- Product: AI role-play for English learning (phone-first + YouTube-based practice)
- Primary users: shy / camera-avoidant learners, busy professionals, audio-first learners
- Core wedge: private, repeatable speaking practice without scheduling pressure
- Status: in active build + iteration (content + UX evolving alongside user feedback)
Leadership
- Defined the product wedge and experience principles (privacy, repeatability, play)
- Owned UX direction across web + mobile interaction patterns
- Developed the content approach (situations, prompts, “micro-practice” rhythm)
- Shaped brand voice and UX writing to reduce anxiety and keep tone supportive
- Co-Founder / Product Designer
- UX + Content Strategy
- IA, Flows, Wireframes, UI
- Brand direction + UX writing
- Figma / FigJam
- Notion (requirements + planning)
- Adobe Illustrator (logo design & graphics)
- Next.js (prototyping & production)
The Problem
Learners don’t fail because they “don’t study” — they fail because speaking feels risky
Speaking practice is the highest-friction part of learning English:
- Video calls create pressure (camera anxiety, fear of judgment).
- Human tutors are costly and hard to scale for daily repetition.
- Most apps stay on-screen, so practice doesn’t feel like real conversation.
If we make speaking feel private, repeatable, and playful, learners will practice more often — and confidence will follow.
- Shy / camera-avoidant learners
- Busy professionals (micro-practice)
- Audio-first learners (commute / chores)
- Learners who copy real speech from content
Product Approach
Two immersive modes, one goal: real speaking practice
Snackk.me combines audio role-play and YouTube role-play so learners can practice in the way that feels most natural — phone-first conversation or content-driven imitation.
AI Audio Role-play (Phone App)
Short, low-pressure speaking sessions designed to feel like a real phone call — ideal for building confidence and habit.
- Shy-learner friendly: no camera, no strangers.
- Micro-practice: 3–10 minute “snack” sessions.
- Role-play scenarios: work, travel, cooking, sports.
YouTube Role-play (Website)
Watch a video, then practice speaking in-context — Snackk extracts learnable moments so users can role-play right next to the content.
- Learn from real speech: phrases, tone, social context.
- Extracted learning prompts: practice without hunting for vocabulary.
- Immersive pairing: video on one side, practice on the other.
Core Features
Play situations, find words, and practice with real content
The experience is built around context — because learners don’t speak in “chapters,” they speak in situations.
Play Situation
Explore by Situation — Select a context to start playing
Quick prompts & common phrases
Practical speaking moments
Casual conversation & slang
Meetings, updates, small talk
This feature is intentionally “low decision cost” — pick a situation and start speaking. It reduces the blank-page anxiety of “What should I talk about?”
Explore Words
Find the Perfect Word — Search by context, meaning, or feeling
Learners often know what they want to express, but not the natural word. Explore Words is built for intent-first search.
- By context: “work meeting”, “restaurant”, “dating”, “small talk”
- By meaning: synonyms that match level + tone
- By feeling: polite, confident, casual, serious
This supports more natural speaking because it maps language to the learner’s real intent, not just dictionary definitions.
YouTube Role-play
Practice with the content you already watch — extracted prompts appear right next to the video
- Content extraction: pull key phrases and speaking moments from the video.
- Guided role-play: learners respond in-context (tone + intent matters).
- On-page practice: the “learning layer” sits beside the video for immersion.
This feature blends entertainment with practice — making it easier for learners to stay consistent.
Phone App
AI audio role-play that feels private and real
The phone experience is designed for shy learners: no camera, quick starts, clear states, and a supportive tone that makes mistakes feel normal.
Designed for anxiety reduction
Calm UI, friendly microcopy, and predictable flows so learners feel safe speaking out loud.
Fast entry into practice
A learner can start a role-play in seconds: choose a situation, then speak.
Role-play prompts that scale
Scenarios provide structure without making conversation feel scripted or robotic.
Feedback loop (placeholder)
After each session: quick recap, vocabulary, and “try this next time” suggestions.
Portfolio Notes
How the product design decisions were made
I used lightweight research and competitive scanning to focus the product on one sharp wedge: immersive speaking practice without the friction of video calls or formal tutoring.
Problem framing
Speaking practice breaks down due to cost, scheduling, and anxiety — not lack of motivation.
Design goal
Reduce pressure and increase repetition: private, playful, short sessions.
Key bet
Context-first UX (“situations”) beats feature-first UX for learners.
AI principle
AI should feel supportive and predictable — not like a novelty demo.
Snackk.me is built to make speaking practice feel doable
If you’re interested in the product, partnership ideas, or the UX behind immersive learning experiences, I’d love to connect.